Showing 1–50 of 96 results for author: Zou, P

PsycoLLM: Enhancing LLM for Psychological Understanding and Evaluation

Authors: **peng Hu, Tengteng Dong, Hui Ma, Peng Zou, Xiao Sun, Meng Wang

Abstract: Mental health has attracted substantial attention in recent years and LLM can be an effective technology for alleviating this problem owing to its capability in text understanding and dialogue. However, existing research in this domain often suffers from limitations, such as training on datasets lacking crucial prior knowledge and evidence, and the absence of comprehensive evaluation methods. In t… ▽ More Mental health has attracted substantial attention in recent years and LLM can be an effective technology for alleviating this problem owing to its capability in text understanding and dialogue. However, existing research in this domain often suffers from limitations, such as training on datasets lacking crucial prior knowledge and evidence, and the absence of comprehensive evaluation methods. In this paper, we propose a specialized psychological large language model (LLM), named PsycoLLM, trained on a proposed high-quality psychological dataset, including single-turn QA, multi-turn dialogues enriched with prior knowledge and knowledge-based QA. Additionally, to compare the performance of PsycoLLM with other LLMs, we develop a comprehensive psychological benchmark based on authoritative psychological counseling examinations in China, which includes assessments of professional ethics, theoretical proficiency, and case analysis. The experimental results on the benchmark illustrates the effectiveness of PsycoLLM, which demonstrates superior performance compared to other LLMs. △ Less

Submitted 8 July, 2024; originally announced July 2024.

Comments: work in progress

Large Language Models Are Involuntary Truth-Tellers: Exploiting Fallacy Failure for Jailbreak Attacks

Authors: Yue Zhou, Henry Peng Zou, Barbara Di Eugenio, Yang Zhang

Abstract: We find that language models have difficulties generating fallacious and deceptive reasoning. When asked to generate deceptive outputs, language models tend to leak honest counterparts but believe them to be false. Exploiting this deficiency, we propose a jailbreak attack method that elicits an aligned language model for malicious output. Specifically, we query the model to generate a fallacious y… ▽ More We find that language models have difficulties generating fallacious and deceptive reasoning. When asked to generate deceptive outputs, language models tend to leak honest counterparts but believe them to be false. Exploiting this deficiency, we propose a jailbreak attack method that elicits an aligned language model for malicious output. Specifically, we query the model to generate a fallacious yet deceptively real procedure for the harmful behavior. Since a fallacious procedure is generally considered fake and thus harmless by LLMs, it helps bypass the safeguard mechanism. Yet the output is factually harmful since the LLM cannot fabricate fallacious solutions but proposes truthful ones. We evaluate our approach over five safety-aligned large language models, comparing four previous jailbreak methods, and show that our approach achieves competitive performance with more harmful outputs. We believe the findings could be extended beyond model safety, such as self-verification and hallucination. △ Less

Submitted 30 June, 2024; originally announced July 2024.

LLMs Assist NLP Researchers: Critique Paper (Meta-)Reviewing

Authors: Jiangshu Du, Yibo Wang, Wenting Zhao, Zhongfen Deng, Shuaiqi Liu, Renze Lou, Henry Peng Zou, Pranav Narayanan Venkit, Nan Zhang, Mukund Srinath, Haoran Ranran Zhang, Vipul Gupta, Yinghui Li, Tao Li, Fei Wang, Qin Liu, Tianlin Liu, Pengzhi Gao, Congying Xia, Chen Xing, Jiayang Cheng, Zhaowei Wang, Ying Su, Raj Sanjay Shah, Ruohao Guo , et al. (15 additional authors not shown)

Abstract: This work is motivated by two key trends. On one hand, large language models (LLMs) have shown remarkable versatility in various generative tasks such as writing, drawing, and question answering, significantly reducing the time required for many routine tasks. On the other hand, researchers, whose work is not only time-consuming but also highly expertise-demanding, face increasing challenges as th… ▽ More This work is motivated by two key trends. On one hand, large language models (LLMs) have shown remarkable versatility in various generative tasks such as writing, drawing, and question answering, significantly reducing the time required for many routine tasks. On the other hand, researchers, whose work is not only time-consuming but also highly expertise-demanding, face increasing challenges as they have to spend more time reading, writing, and reviewing papers. This raises the question: how can LLMs potentially assist researchers in alleviating their heavy workload? This study focuses on the topic of LLMs assist NLP Researchers, particularly examining the effectiveness of LLM in assisting paper (meta-)reviewing and its recognizability. To address this, we constructed the ReviewCritique dataset, which includes two types of information: (i) NLP papers (initial submissions rather than camera-ready) with both human-written and LLM-generated reviews, and (ii) each review comes with "deficiency" labels and corresponding explanations for individual segments, annotated by experts. Using ReviewCritique, this study explores two threads of research questions: (i) "LLMs as Reviewers", how do reviews generated by LLMs compare with those written by humans in terms of quality and distinguishability? (ii) "LLMs as Metareviewers", how effectively can LLMs identify potential issues, such as Deficient or unprofessional review segments, within individual paper reviews? To our knowledge, this is the first work to provide such a comprehensive analysis. △ Less

Submitted 25 June, 2024; v1 submitted 23 June, 2024; originally announced June 2024.

Interstellar Nitrogen Isotope Ratios: Measurements on tracers of C$^{14}$N and C$^{15}$N

Authors: J. L. Chen, J. S. Zhang, C. Henkel, Y. T. Yan, H. Z. Yu, Y. X. Wang, Y. P. Zou, J. Y. Zhao, X. Y. Wang

Abstract: The nitrogen isotope ratio 14N/15N is a powerful tool to trace Galactic stellar nucleosynthesis and constraining Galactic chemical evolution. Previous observations have found lower 14N/15N ratios in the Galactic center and higher values in the Galactic disk. This is consistent with the inside-out formation scenario of our Milky Way. However, previous studies mostly utilized double isotope ratios a… ▽ More The nitrogen isotope ratio 14N/15N is a powerful tool to trace Galactic stellar nucleosynthesis and constraining Galactic chemical evolution. Previous observations have found lower 14N/15N ratios in the Galactic center and higher values in the Galactic disk. This is consistent with the inside-out formation scenario of our Milky Way. However, previous studies mostly utilized double isotope ratios also including 12C/13C, which introduces additional uncertainties. Here we therefore present observations of C14N and its rare isotopologue, C15N, toward a sample of star forming regions, measured by the IRAM 30 m and/or the ARO 12 m telescope at $λ$ ~3 mm wavelength. For those 35 sources detected in both isotopologues, physical parameters are determined. Furthermore we have obtained nitrogen isotope ratios using the strongest hyperfine components of CN and C15N. For those sources showing small deviations from Local Thermodynamical Equilibrium and/or self-absorption, the weakest hyperfine component, likely free of the latter effect, was used to obtain reliable 14N/15N values. Our measured 14N/15N isotope ratios from C14N and C15N measurements are compatible with those from our earlier measurements of NH3 and 15NH3 (Paper I), i.e., increasing ratios to a Galacticentric distance of ~9 kpc. The unweighted second order polynomial fit yields $\frac{{\rm C^{14}N}}{{\rm C^{15}N}} = (-4.85 \pm 1.89)\;{\rm kpc^{-2}} \times R_{\rm GC}^{2} + (82.11 \pm 31.93) \;{\rm kpc^{-1}} \times R_{\rm GC} - (28.12 \pm 126.62)$. Toward the outer galaxy, the isotope ratio tends to decrease, supporting an earlier finding by H13CN/HC15N. Galactic chemical evolution models are consistent with our measurements of the 14N/15N isotope ratio, i.e. a rising trend from the Galactic center region to approximately 9 kpc, followed by a decreasing trend with increasing $R_{\rm GC}$ toward the outer Galaxy. △ Less

Submitted 13 June, 2024; originally announced June 2024.

Comments: 34 pages, 9 figures, 6 tables

Journal ref: The Astrophysical Journal (2004)

Deconstructing The Ethics of Large Language Models from Long-standing Issues to New-emerging Dilemmas

Authors: Chengyuan Deng, Yiqun Duan, Xin **, Heng Chang, Yijun Tian, Han Liu, Henry Peng Zou, Yiqiao **, Yijia Xiao, Yichen Wang, Shenghao Wu, Zongxing Xie, Kuofeng Gao, Sihong He, Jun Zhuang, Lu Cheng, Haohan Wang

Abstract: Large Language Models (LLMs) have achieved unparalleled success across diverse language modeling tasks in recent years. However, this progress has also intensified ethical concerns, impacting the deployment of LLMs in everyday contexts. This paper provides a comprehensive survey of ethical challenges associated with LLMs, from longstanding issues such as copyright infringement, systematic bias, an… ▽ More Large Language Models (LLMs) have achieved unparalleled success across diverse language modeling tasks in recent years. However, this progress has also intensified ethical concerns, impacting the deployment of LLMs in everyday contexts. This paper provides a comprehensive survey of ethical challenges associated with LLMs, from longstanding issues such as copyright infringement, systematic bias, and data privacy, to emerging problems like truthfulness and social norms. We critically analyze existing research aimed at understanding, examining, and mitigating these ethical risks. Our survey underscores integrating ethical standards and societal values into the development of LLMs, thereby guiding the development of responsible and ethically aligned language models. △ Less

Submitted 8 June, 2024; originally announced June 2024.

Mixed Supervised Graph Contrastive Learning for Recommendation

Authors: Weizhi Zhang, Liangwei Yang, Zihe Song, Henry Peng Zou, Ke Xu, Yuanjie Zhu, Philip S. Yu

Abstract: Recommender systems (RecSys) play a vital role in online platforms, offering users personalized suggestions amidst vast information. Graph contrastive learning aims to learn from high-order collaborative filtering signals with unsupervised augmentation on the user-item bipartite graph, which predominantly relies on the multi-task learning framework involving both the pair-wise recommendation loss… ▽ More Recommender systems (RecSys) play a vital role in online platforms, offering users personalized suggestions amidst vast information. Graph contrastive learning aims to learn from high-order collaborative filtering signals with unsupervised augmentation on the user-item bipartite graph, which predominantly relies on the multi-task learning framework involving both the pair-wise recommendation loss and the contrastive loss. This decoupled design can cause inconsistent optimization direction from different losses, which leads to longer convergence time and even sub-optimal performance. Besides, the self-supervised contrastive loss falls short in alleviating the data sparsity issue in RecSys as it learns to differentiate users/items from different views without providing extra supervised collaborative filtering signals during augmentations. In this paper, we propose Mixed Supervised Graph Contrastive Learning for Recommendation (MixSGCL) to address these concerns. MixSGCL originally integrates the training of recommendation and unsupervised contrastive losses into a supervised contrastive learning loss to align the two tasks within one optimization direction. To cope with the data sparsity issue, instead unsupervised augmentation, we further propose node-wise and edge-wise mixup to mine more direct supervised collaborative filtering signals based on existing user-item interactions. Extensive experiments on three real-world datasets demonstrate that MixSGCL surpasses state-of-the-art methods, achieving top performance on both accuracy and efficiency. It validates the effectiveness of MixSGCL with our coupled design on supervised graph contrastive learning. △ Less

Submitted 25 April, 2024; v1 submitted 24 April, 2024; originally announced April 2024.

ImplicitAVE: An Open-Source Dataset and Multimodal LLMs Benchmark for Implicit Attribute Value Extraction

Authors: Henry Peng Zou, Vinay Samuel, Yue Zhou, Weizhi Zhang, Liancheng Fang, Zihe Song, Philip S. Yu, Cornelia Caragea

Abstract: Existing datasets for attribute value extraction (AVE) predominantly focus on explicit attribute values while neglecting the implicit ones, lack product images, are often not publicly available, and lack an in-depth human inspection across diverse domains. To address these limitations, we present ImplicitAVE, the first, publicly available multimodal dataset for implicit attribute value extraction.… ▽ More Existing datasets for attribute value extraction (AVE) predominantly focus on explicit attribute values while neglecting the implicit ones, lack product images, are often not publicly available, and lack an in-depth human inspection across diverse domains. To address these limitations, we present ImplicitAVE, the first, publicly available multimodal dataset for implicit attribute value extraction. ImplicitAVE, sourced from the MAVE dataset, is carefully curated and expanded to include implicit AVE and multimodality, resulting in a refined dataset of 68k training and 1.6k testing data across five domains. We also explore the application of multimodal large language models (MLLMs) to implicit AVE, establishing a comprehensive benchmark for MLLMs on the ImplicitAVE dataset. Six recent MLLMs with eleven variants are evaluated across diverse settings, revealing that implicit value extraction remains a challenging task for MLLMs. The contributions of this work include the development and release of ImplicitAVE, and the exploration and benchmarking of various MLLMs for implicit AVE, providing valuable insights and potential future research directions. Dataset and code are available at https://github.com/HenryPengZou/ImplicitAVE △ Less

Submitted 23 April, 2024; originally announced April 2024.

EIVEN: Efficient Implicit Attribute Value Extraction using Multimodal LLM

Authors: Henry Peng Zou, Gavin Heqing Yu, Ziwei Fan, Dan Bu, Han Liu, Peng Dai, Dongmei Jia, Cornelia Caragea

Abstract: In e-commerce, accurately extracting product attribute values from multimodal data is crucial for improving user experience and operational efficiency of retailers. However, previous approaches to multimodal attribute value extraction often struggle with implicit attribute values embedded in images or text, rely heavily on extensive labeled data, and can easily confuse similar attribute values. To… ▽ More In e-commerce, accurately extracting product attribute values from multimodal data is crucial for improving user experience and operational efficiency of retailers. However, previous approaches to multimodal attribute value extraction often struggle with implicit attribute values embedded in images or text, rely heavily on extensive labeled data, and can easily confuse similar attribute values. To address these issues, we introduce EIVEN, a data- and parameter-efficient generative framework that pioneers the use of multimodal LLM for implicit attribute value extraction. EIVEN leverages the rich inherent knowledge of a pre-trained LLM and vision encoder to reduce reliance on labeled data. We also introduce a novel Learning-by-Comparison technique to reduce model confusion by enforcing attribute value comparison and difference identification. Additionally, we construct initial open-source datasets for multimodal implicit attribute value extraction. Our extensive experiments reveal that EIVEN significantly outperforms existing methods in extracting implicit attribute values while requiring less labeled data. △ Less

Submitted 12 April, 2024; originally announced April 2024.

Comments: Accepted by NAACL 2024 Industry Track

Age of Information Optimization and State Error Analysis for Correlated Multi-Process Multi-Sensor Systems

Authors: Egemen Erbayat, Ali Maatouk, Peng Zou, Suresh Subramaniam

Abstract: In this paper, we examine a multi-sensor system where each sensor may monitor more than one time-varying information process and send status updates to a remote monitor over a common channel. We consider that each sensor's status update may contain information about more than one information process in the system subject to the system's constraints. To investigate the impact of this correlation on… ▽ More In this paper, we examine a multi-sensor system where each sensor may monitor more than one time-varying information process and send status updates to a remote monitor over a common channel. We consider that each sensor's status update may contain information about more than one information process in the system subject to the system's constraints. To investigate the impact of this correlation on the overall system's performance, we conduct an analysis of both the average Age of Information (AoI) and source state estimation error at the monitor. Building upon this analysis, we subsequently explore the impact of the packet arrivals, correlation probabilities, and rate of processes' state change on the system's performance. Next, we consider the case where sensors have limited sensing abilities and distribute a portion of their sensing abilities across the different processes. We optimize this distribution to minimize the total AoI of the system. Interestingly, we show that monitoring multiple processes from a single source may not always be beneficial. Our results also reveal that the optimal sensing distribution for diverse arrival rates may exhibit a rapid regime switch, rather than smooth transitions, after crossing critical system values. This highlights the importance of identifying these critical thresholds to ensure effective system performance. △ Less

Submitted 2 July, 2024; v1 submitted 12 April, 2024; originally announced April 2024.

Comments: fix typos

ByteComposer: a Human-like Melody Composition Method based on Language Model Agent

Authors: Xia Liang, Xingjian Du, Jiaju Lin, Pei Zou, Yuan Wan, Bilei Zhu

Abstract: Large Language Models (LLM) have shown encouraging progress in multimodal understanding and generation tasks. However, how to design a human-aligned and interpretable melody composition system is still under-explored. To solve this problem, we propose ByteComposer, an agent framework emulating a human's creative pipeline in four separate steps : "Conception Analysis - Draft Composition - Self-Eval… ▽ More Large Language Models (LLM) have shown encouraging progress in multimodal understanding and generation tasks. However, how to design a human-aligned and interpretable melody composition system is still under-explored. To solve this problem, we propose ByteComposer, an agent framework emulating a human's creative pipeline in four separate steps : "Conception Analysis - Draft Composition - Self-Evaluation and Modification - Aesthetic Selection". This framework seamlessly blends the interactive and knowledge-understanding features of LLMs with existing symbolic music generation models, thereby achieving a melody composition agent comparable to human creators. We conduct extensive experiments on GPT4 and several open-source large language models, which substantiate our framework's effectiveness. Furthermore, professional music composers were engaged in multi-dimensional evaluations, the final results demonstrated that across various facets of music composition, ByteComposer agent attains the level of a novice melody composer. △ Less

Submitted 6 March, 2024; v1 submitted 23 February, 2024; originally announced February 2024.

A universal pairing gap measurement proposal by dynamical excitations in 2D doped attractive Fermi-Hubbard model with spin-orbit coupling

Authors: Huaisong Zhao, Rui Han, Ling Qin, Feng Yuan, Peng Zou

Abstract: By calculating dynamical structure factor of two-dimensional doped attractive Fermi-Hubbard model with Rashba spin-orbit coupling, we not only investigate collective modes and single-particle excitations of the system during the phase transition between Bardeen-Cooper-Schrieffer superfluid and topological superfluid, but also propose a universal method to measure pairing gap measurement in an opti… ▽ More By calculating dynamical structure factor of two-dimensional doped attractive Fermi-Hubbard model with Rashba spin-orbit coupling, we not only investigate collective modes and single-particle excitations of the system during the phase transition between Bardeen-Cooper-Schrieffer superfluid and topological superfluid, but also propose a universal method to measure pairing gap measurement in an optical lattice system. Our numerical results show that the area of the molecular excitation peak at the transferred momentum ${\bf q}=\left[π,π\right]$ is proportional to the square of the pairing gap in the system with Rashba SOC. In particular, this method is very sensitive to the pairing gap. This goes on verifying that this method is universal to measure the pairing gap in a doped optical lattice with Rashba SOC. These theoretical results are important for experimentally measuring the pairing gap and studying the topological superfluid in an optical lattice. △ Less

Submitted 30 January, 2024; originally announced January 2024.

Comments: 12 pages, 8 figures

Dynamical excitations of one-dimensional Fulde-Ferrell pairing Fermi superfluid

Authors: Peng Zou, Huaisong Zhao, Feng Yuan, Shi-Guo Peng

Abstract: We theoretically investigate a one-dimensional Fulde-Ferrell Fermi superfluid at a finite effective Zeeman field $h$, and study entire dynamical excitations related to density perturbation. By calculating the density dynamic structure factor, we find anisotropic dynamical excitations in both collective modes and single-particle excitations. Along the direction of centre-of-mass momentum $p$, there… ▽ More We theoretically investigate a one-dimensional Fulde-Ferrell Fermi superfluid at a finite effective Zeeman field $h$, and study entire dynamical excitations related to density perturbation. By calculating the density dynamic structure factor, we find anisotropic dynamical excitations in both collective modes and single-particle excitations. Along the direction of centre-of-mass momentum $p$, there are two obvious gapless collective modes with different speed. The lower collective modes is from the usual gauge symmetry breaking and has a larger speed than the one in the negative direction of $p$. The higher one is due to the direction spontaneous symmetry breaking of centre-of-mass momentum $p$, and separates two kinds of single-particle excitations in the positive $p$ direction. However, this higher mode disappears in the opposite direction of $p$, where two single-particle excitations overlap with each other. These signals of dynamical excitations can do help to distinguish Fulde-Ferrell superfluid from the conventional Bardeen-Cooper-Schrieffer superfluid in the future experiment. △ Less

Submitted 4 January, 2024; originally announced January 2024.

Comments: 7 pages, 5 figures

Multiscale Quantum Approximate Optimization Algorithm

Authors: ** Zou

Abstract: The quantum approximate optimization algorithm (QAOA) is one of the canonical algorithms designed to find approximate solutions to combinatorial optimization problems in current noisy intermediate-scale quantum (NISQ) devices. It is an active area of research to exhibit its speedup over classical algorithms. The performance of the QAOA at low depths is limited, while the QAOA at higher depths is c… ▽ More The quantum approximate optimization algorithm (QAOA) is one of the canonical algorithms designed to find approximate solutions to combinatorial optimization problems in current noisy intermediate-scale quantum (NISQ) devices. It is an active area of research to exhibit its speedup over classical algorithms. The performance of the QAOA at low depths is limited, while the QAOA at higher depths is constrained by the current techniques. We propose a new version of QAOA that incorporates the capabilities of QAOA and the real-space renormalization group transformation, resulting in enhanced performance. Numerical simulations demonstrate that our algorithm can provide accurate solutions for certain randomly generated instances utilizing QAOA at low depths, even at the lowest depth. The algorithm is suitable for NISQ devices to exhibit a quantum advantage. △ Less

Submitted 11 December, 2023; originally announced December 2023.

A Systematic Observational Study on Galactic Interstellar Ratio 18O/17O. II. C18O and C17O J=2-1 Data Analysis

Authors: Y. P. Zou, J. S. Zhang, C. Henkel, D. Romano, W. Liu, Y. H. Zheng, Y. T. Yan, J. L. Chen, Y. X. Wang, J. Y. Zhao

Abstract: To investigate the relative amount of ejecta from high-mass versus intermediate-mass stars and to trace the chemical evolution of the Galaxy, we have performed with the IRAM 30m and the SMT 10m telescopes a systematic study of Galactic interstellar 18O/17O ratios toward a sample of 421 molecular clouds, covering a galactocentric distance range of 1-22 kpc. The results presented in this paper are b… ▽ More To investigate the relative amount of ejecta from high-mass versus intermediate-mass stars and to trace the chemical evolution of the Galaxy, we have performed with the IRAM 30m and the SMT 10m telescopes a systematic study of Galactic interstellar 18O/17O ratios toward a sample of 421 molecular clouds, covering a galactocentric distance range of 1-22 kpc. The results presented in this paper are based on the J=2-1 transition and encompass 364 sources showing both C18O and C17O detections. The previously suggested 18O/17O gradient is confirmed. For the 41 sources detected with both facilities, good agreement is obtained. A correlation of 18O/17O ratios with heliocentric distance is not found, indicating that beam dilution and linear beam sizes are not relevant. For the subsample of IRAM 30 m high-mass star-forming regions with accurate parallax distances, an unweighted fit gives 18O/17O = (0.12+-0.02)R_GC+(2.38+-0.13) with a correlation coefficient of R = 0.67. While the slope is consistent with our J=1-0 measurement, ratios are systematically lower. This should be caused by larger optical depths of C18O 2-1 lines, w.r.t the corresponding 1-0 transitions, which is supported by RADEX calculations and the fact that C18O/C17O is positively correlated with 13CO/C18O. After considering optical depth effects with C18O J=2-1 reaching typically an optical depth of 0.5, corrected 18O/17O ratios from the J=1-0 and J=2-1 lines become consistent. A good numerical fit to the data is provided by the MWG-12 model, including both rotating stars and novae. △ Less

Submitted 1 November, 2023; originally announced November 2023.

Comments: 17 pages, 11 figures, published in ApJS

CrisisMatch: Semi-Supervised Few-Shot Learning for Fine-Grained Disaster Tweet Classification

Authors: Henry Peng Zou, Yue Zhou, Cornelia Caragea, Doina Caragea

Abstract: The shared real-time information about natural disasters on social media platforms like Twitter and Facebook plays a critical role in informing volunteers, emergency managers, and response organizations. However, supervised learning models for monitoring disaster events require large amounts of annotated data, making them unrealistic for real-time use in disaster events. To address this challenge,… ▽ More The shared real-time information about natural disasters on social media platforms like Twitter and Facebook plays a critical role in informing volunteers, emergency managers, and response organizations. However, supervised learning models for monitoring disaster events require large amounts of annotated data, making them unrealistic for real-time use in disaster events. To address this challenge, we present a fine-grained disaster tweet classification model under the semi-supervised, few-shot learning setting where only a small number of annotated data is required. Our model, CrisisMatch, effectively classifies tweets into fine-grained classes of interest using few labeled data and large amounts of unlabeled data, mimicking the early stage of a disaster. Through integrating effective semi-supervised learning ideas and incorporating TextMixUp, CrisisMatch achieves performance improvement on two disaster datasets of 11.2\% on average. Further analyses are also provided for the influence of the number of labeled data and out-of-domain results. △ Less

Submitted 23 October, 2023; originally announced October 2023.

Comments: Accepted by ISCRAM 2023

JointMatch: A Unified Approach for Diverse and Collaborative Pseudo-Labeling to Semi-Supervised Text Classification

Authors: Henry Peng Zou, Cornelia Caragea

Abstract: Semi-supervised text classification (SSTC) has gained increasing attention due to its ability to leverage unlabeled data. However, existing approaches based on pseudo-labeling suffer from the issues of pseudo-label bias and error accumulation. In this paper, we propose JointMatch, a holistic approach for SSTC that addresses these challenges by unifying ideas from recent semi-supervised learning an… ▽ More Semi-supervised text classification (SSTC) has gained increasing attention due to its ability to leverage unlabeled data. However, existing approaches based on pseudo-labeling suffer from the issues of pseudo-label bias and error accumulation. In this paper, we propose JointMatch, a holistic approach for SSTC that addresses these challenges by unifying ideas from recent semi-supervised learning and the task of learning with noise. JointMatch adaptively adjusts classwise thresholds based on the learning status of different classes to mitigate model bias towards current easy classes. Additionally, JointMatch alleviates error accumulation by utilizing two differently initialized networks to teach each other in a cross-labeling manner. To maintain divergence between the two networks for mutual learning, we introduce a strategy that weighs more disagreement data while also allowing the utilization of high-quality agreement data for training. Experimental results on benchmark datasets demonstrate the superior performance of JointMatch, achieving a significant 5.13% improvement on average. Notably, JointMatch delivers impressive results even in the extremely-scarce-label setting, obtaining 86% accuracy on AG News with only 5 labels per class. We make our code available at https://github.com/HenryPengZou/JointMatch. △ Less

Submitted 23 October, 2023; originally announced October 2023.

Comments: Accepted by EMNLP 2023 (Main)

DeCrisisMB: Debiased Semi-Supervised Learning for Crisis Tweet Classification via Memory Bank

Authors: Henry Peng Zou, Yue Zhou, Weizhi Zhang, Cornelia Caragea

Abstract: During crisis events, people often use social media platforms such as Twitter to disseminate information about the situation, warnings, advice, and support. Emergency relief organizations leverage such information to acquire timely crisis circumstances and expedite rescue operations. While existing works utilize such information to build models for crisis event analysis, fully-supervised approache… ▽ More During crisis events, people often use social media platforms such as Twitter to disseminate information about the situation, warnings, advice, and support. Emergency relief organizations leverage such information to acquire timely crisis circumstances and expedite rescue operations. While existing works utilize such information to build models for crisis event analysis, fully-supervised approaches require annotating vast amounts of data and are impractical due to limited response time. On the other hand, semi-supervised models can be biased, performing moderately well for certain classes while performing extremely poorly for others, resulting in substantially negative effects on disaster monitoring and rescue. In this paper, we first study two recent debiasing methods on semi-supervised crisis tweet classification. Then we propose a simple but effective debiasing method, DeCrisisMB, that utilizes a Memory Bank to store and perform equal sampling for generated pseudo-labels from each class at each training iteration. Extensive experiments are conducted to compare different debiasing methods' performance and generalization ability in both in-distribution and out-of-distribution settings. The results demonstrate the superior performance of our proposed method. Our code is available at https://github.com/HenryPengZou/DeCrisisMB. △ Less

Submitted 23 October, 2023; originally announced October 2023.

Comments: Accepted by EMNLP 2023 (Findings)

A ModelOps-based Framework for Intelligent Medical Knowledge Extraction

Authors: Hongxin Ding, Peinie Zou, Zhiyuan Wang, Junfeng Zhao, Yasha Wang, Qiang Zhou

Abstract: Extracting medical knowledge from healthcare texts enhances downstream tasks like medical knowledge graph construction and clinical decision-making. However, the construction and application of knowledge extraction models lack automation, reusability and unified management, leading to inefficiencies for researchers and high barriers for non-AI experts such as doctors, to utilize knowledge extracti… ▽ More Extracting medical knowledge from healthcare texts enhances downstream tasks like medical knowledge graph construction and clinical decision-making. However, the construction and application of knowledge extraction models lack automation, reusability and unified management, leading to inefficiencies for researchers and high barriers for non-AI experts such as doctors, to utilize knowledge extraction. To address these issues, we propose a ModelOps-based intelligent medical knowledge extraction framework that offers a low-code system for model selection, training, evaluation and optimization. Specifically, the framework includes a dataset abstraction mechanism based on multi-layer callback functions, a reusable model training, monitoring and management mechanism. We also propose a model recommendation method based on dataset similarity, which helps users quickly find potentially suitable models for a given dataset. Our framework provides convenience for researchers to develop models and simplifies model access for non-AI experts such as doctors. △ Less

Submitted 4 October, 2023; originally announced October 2023.

PP-MeT: a Real-world Personalized Prompt based Meeting Transcription System

Authors: Xiang Lyu, Yuhang Cao, Qing Wang, **g**g Yin, Yuguang Yang, Pengpeng Zou, Yanni Hu, Heng Lu

Abstract: Speaker-attributed automatic speech recognition (SA-ASR) improves the accuracy and applicability of multi-speaker ASR systems in real-world scenarios by assigning speaker labels to transcribed texts. However, SA-ASR poses unique challenges due to factors such as speaker overlap, speaker variability, background noise, and reverberation. In this study, we propose PP-MeT system, a real-world personal… ▽ More Speaker-attributed automatic speech recognition (SA-ASR) improves the accuracy and applicability of multi-speaker ASR systems in real-world scenarios by assigning speaker labels to transcribed texts. However, SA-ASR poses unique challenges due to factors such as speaker overlap, speaker variability, background noise, and reverberation. In this study, we propose PP-MeT system, a real-world personalized prompt based meeting transcription system, which consists of a clustering system, target-speaker voice activity detection (TS-VAD), and TS-ASR. Specifically, we utilize target-speaker embedding as a prompt in TS-VAD and TS-ASR modules in our proposed system. In constrast with previous system, we fully leverage pre-trained models for system initialization, thereby bestowing our approach with heightened generalizability and precision. Experiments on M2MeT2.0 Challenge dataset show that our system achieves a cp-CER of 11.27% on the test set, ranking first in both fixed and open training conditions. △ Less

Submitted 28 September, 2023; originally announced September 2023.

Magnetic Reconnection as the Key Mechanism in Sunspot Rotation Leading to Solar Eruption

Authors: Chaowei Jiang, Xueshang Feng, Xinkai Bian, Peng Zou, Aiying Duan, Xiaoli Yan, Qiang Hu, Wen He, Xinyi Wang, **bing Zuo, Yi Wang

Abstract: The rotation of sunspots around their umbral center has long been considered as an important process in leading to solar eruptions, but the underlying mechanism remains unclear. A prevailing physical picture on how sunspot rotation leads to eruption is that, by twisting the coronal magnetic field lines from their footpoints, the rotation can build up a magnetic flux rope and drive it into some kin… ▽ More The rotation of sunspots around their umbral center has long been considered as an important process in leading to solar eruptions, but the underlying mechanism remains unclear. A prevailing physical picture on how sunspot rotation leads to eruption is that, by twisting the coronal magnetic field lines from their footpoints, the rotation can build up a magnetic flux rope and drive it into some kinds of ideal magnetohydrodynamics (MHD) instabilities which initiate eruptions. Here with a data-inspired MHD simulation we studied the rotation of a large sunspot in solar active region NOAA 12158 leading to a major eruption, and found that it is distinct from prevailing theories based on ideal instabilities of twisted flux rope. The simulation suggests that, through successive rotation of the sunspot, the coronal magnetic field is sheared with a central current sheet created progressively within the sheared arcade before the eruption, but without forming a flux rope. Then the eruption is instantly triggered once fast reconnection sets in at the current sheet, while a highly twisted flux rope is created during the eruption. Furthermore, the simulation reveals an intermediate evolution stage between the quasi-static energy-storage phase and the impulsive eruption-acceleration phase. This stage may correspond to the slow-rise phase in observation and it enhances building up of the current sheet. △ Less

Submitted 30 September, 2023; v1 submitted 19 August, 2023; originally announced August 2023.

Comments: Updated from the initial version and text overlap with arXiv:2308.06977 is removed

Data-driven MHD simulation of a sunspot rotating active region leading to solar eruption

Authors: Chaowei Jiang, Xueshang Feng, Xinkai Bian, Peng Zou, Aiying Duan, Xiaoli Yan, Qiang Hu, Wen He, Xinyi Wang, **bing Zuo, Yi Wang

Abstract: Solar eruptions are the leading driver of space weather, and it is vital for space weather forecast to understand in what conditions the solar eruptions can be produced and how they are initiated. The rotation of sunspots around their umbral center has long been considered as an important condition in causing solar eruptions. To unveil the underlying mechanisms, here we carried out a data-driven m… ▽ More Solar eruptions are the leading driver of space weather, and it is vital for space weather forecast to understand in what conditions the solar eruptions can be produced and how they are initiated. The rotation of sunspots around their umbral center has long been considered as an important condition in causing solar eruptions. To unveil the underlying mechanisms, here we carried out a data-driven magnetohydrodynamics simulation for the event of a large sunspot with rotation for days in solar active region NOAA 12158 leading to a major eruption. The photospheric velocity as recovered from the time sequence of vector magnetograms are inputted directly at the bottom boundary of the numerical model as the driving flow. Our simulation successfully follows the long-term quasi-static evolution of the active region until the fast eruption, with magnetic field structure consistent with the observed coronal emission and onset time of simulated eruption matches rather well with the observations. Analysis of the process suggests that through the successive rotation of the sunspot the coronal magnetic field is sheared with a vertical current sheet created progressively, and once fast reconnection sets in at the current sheet, the eruption is instantly triggered, with a highly twisted flux rope originating from the eruption. This data-driven simulation stresses magnetic reconnection as the key mechanism in sunspot rotation leading to eruption. △ Less

Submitted 14 August, 2023; originally announced August 2023.

Comments: Accept by A&A

A model of failed solar eruption initiated and destructed by magnetic reconnection

Authors: Chaowei Jiang, Aiying Duan, Peng Zou, Zhenjun Zhou, Xinkai Bian, Xueshang Feng, **bing Zuo, Yi Wang

Abstract: Solar eruptions are explosive disruption of coronal magnetic fields, and often launch coronal mass ejections into the interplanetary space. Intriguingly, many solar eruptions fail to escape from the Sun, and the prevailing theory for such failed eruption is based on ideal MHD instabilities of magnetic flux rope (MFR); that is, a MFR runs into kink instability and erupts but cannot reach the height… ▽ More Solar eruptions are explosive disruption of coronal magnetic fields, and often launch coronal mass ejections into the interplanetary space. Intriguingly, many solar eruptions fail to escape from the Sun, and the prevailing theory for such failed eruption is based on ideal MHD instabilities of magnetic flux rope (MFR); that is, a MFR runs into kink instability and erupts but cannot reach the height for torus instability. Here, based on numerical MHD simulation, we present a new model of failed eruption in which magnetic reconnection plays a leading role in the initiation and failure of the eruption. Initially, a core bipolar potential field is embedded in a background bipolar field, and by applying shearing and converging motions to the core field, a current sheet is formed within the core field. Then, tether-cutting reconnection is triggered at the current sheet, first slow for a while and becoming fast, driving an erupting MFR. Eventually, the rise of MFR is halted by the downward magnetic tension force of the overlying field, although the MFR apex has well exceeded the critical height of torus instability. More importantly, during the rise of the MFR, it experiences a significant rotation around the vertical axis (with a direction contrary to that predicted by kink instability), rendering the field direction at the rope apex almost inverse to the overlying field. As a result, a strong current sheet is formed between the MFR and the overlying flux, and reconnection occurring in this current sheet ruins completely the MFR. △ Less

Submitted 28 July, 2023; originally announced July 2023.

Comments: Submitted to MNRAS

Dynamic structure factor of two-dimensional Fermi superfluid with Rashba spin-orbit coupling

Authors: Huaisong Zhao, Xu Yan, Shi-Guo Peng, Peng Zou

Abstract: We theoretically calculate the dynamic structure factor of two-dimensional Rashba-type spinorbit coupled (SOC) Fermi superfluid with random phase approximation, and analyse the main characters of dynamical excitation sh own by both density and spin dynamic structure factor during a continuous phase transition between Bardeen-Cooper-Schrieffer superfluid and topological superfluid. Generally we fin… ▽ More We theoretically calculate the dynamic structure factor of two-dimensional Rashba-type spinorbit coupled (SOC) Fermi superfluid with random phase approximation, and analyse the main characters of dynamical excitation sh own by both density and spin dynamic structure factor during a continuous phase transition between Bardeen-Cooper-Schrieffer superfluid and topological superfluid. Generally we find three different excitations, including collective phonon excitation, two-atom molecular and atomic excitations, and pair-breaking excitations due to two-branch structure of quasi-particle spectrum. It should be emphasized that collective phonon excitation is overlapped with a gapless DD type pair-breaking excitation at the critical Zeeman field hc, and is imparted a finite width to phonon peak when transferred momentum q is around Fermi vector kF. At a much larger transferred momentum (q = 4kF ), the pair-breaking excitation happens earlier than two-atom molecular excitation, which is different from the conventional Fermi superfluid without SOC effect. △ Less

Submitted 25 December, 2023; v1 submitted 9 June, 2023; originally announced June 2023.

Comments: 10 pages, 8 figures. arXiv admin note: text overlap with arXiv:2210.10407

Journal ref: Phy. Rev. A 108, 033309 (2023)

Dynamical structure factor and a new method to measure the pairing gap in two-dimensional attractive Fermi-Hubbard model

Authors: Huaisong Zhao, Peng Zou, Feng Yuan

Abstract: By calculating the dynamical structure factor along the high symmetry directions in the Brillouin zone, the dynamical excitations of attractive Fermi-Hubbard model in a two-dimensional square optical lattice are studied with random phase approximation. {Two kinds of collective modes are investigated, including a Goldstone phonon mode at transferred momentum ${\bf q}=\left[0,0\right]$ and a roton m… ▽ More By calculating the dynamical structure factor along the high symmetry directions in the Brillouin zone, the dynamical excitations of attractive Fermi-Hubbard model in a two-dimensional square optical lattice are studied with random phase approximation. {Two kinds of collective modes are investigated, including a Goldstone phonon mode at transferred momentum ${\bf q}=\left[0,0\right]$ and a roton mode at ${\bf q}=\left[π,π\right]$. The phonon origins from the spontaneously U(1) symmetry breaking of pairing gap, and its speed is suppressed by the interaction strength. The collective roton mode origins from the breaking of a global pseudospin SU(2) symmetry.} Dynamical excitations at ${\bf q}=\left[π,π\right]$ consist of a sharp roton molecular peak in the low-energy region and a broad atomic excitation band in the higher energy region. Furthermore, the weight of the roton molecular peak decreases monotonically with increasing the hop** strength, while the weight of the atomic excitations increases quickly. Interestingly we check that the area covered by the roton molecular peak scales with the square of the pairing gap, which is also true in the system with spin-orbit coupling. This conclusion paves a potential way to measure the pairing gap of lattice system experimentally by measuring the dynamical structure factor at ${\bf q}=\left[π,π\right]$. △ Less

Submitted 16 April, 2024; v1 submitted 15 May, 2023; originally announced May 2023.

Comments: 12 pages, 9 figures

How Costly Was That (In)Decision?

Authors: Peng Zou, Ali Maatouk, ** Zhang, Suresh Subramaniam

Abstract: In this paper, we introduce a new metric, named Penalty upon Decision (PuD), for measuring the impact of communication delays and state changes at the source on a remote decision maker. Specifically, the metric quantifies the performance degradation at the decision maker's side due to delayed, erroneous, and (possibly) missed decisions. We clarify the rationale for the metric and derive closed-for… ▽ More In this paper, we introduce a new metric, named Penalty upon Decision (PuD), for measuring the impact of communication delays and state changes at the source on a remote decision maker. Specifically, the metric quantifies the performance degradation at the decision maker's side due to delayed, erroneous, and (possibly) missed decisions. We clarify the rationale for the metric and derive closed-form expressions for its average in M/GI/1 and M/GI/1/1 with blocking settings. Numerical results are then presented to support our expressions and to compare the infinite and zero buffer regimes. Interestingly, comparing these two settings sheds light on a buffer length design challenge that is essential to minimize the average PuD. △ Less

Submitted 24 April, 2023; originally announced April 2023.

Spatial-temporal Transformer for Affective Behavior Analysis

Authors: Peng Zou, Rui Wang, Kehua Wen, Yasi Peng, Xiao Sun

Abstract: The in-the-wild affective behavior analysis has been an important study. In this paper, we submit our solutions for the 5th Workshop and Competition on Affective Behavior Analysis in-the-wild (ABAW), which includes V-A Estimation, Facial Expression Classification and AU Detection Sub-challenges. We propose a Transformer Encoder with Multi-Head Attention framework to learn the distribution of both… ▽ More The in-the-wild affective behavior analysis has been an important study. In this paper, we submit our solutions for the 5th Workshop and Competition on Affective Behavior Analysis in-the-wild (ABAW), which includes V-A Estimation, Facial Expression Classification and AU Detection Sub-challenges. We propose a Transformer Encoder with Multi-Head Attention framework to learn the distribution of both the spatial and temporal features. Besides, there are virious effective data augmentation strategies employed to alleviate the problems of sample imbalance during model training. The results fully demonstrate the effectiveness of our proposed model based on the Aff-Wild2 dataset. △ Less

Submitted 19 March, 2023; originally announced March 2023.

A Possible Chemical Clock in High-mass Star-forming Regions: N(HC3N)/N(N2H+)?

Authors: Y. X. Wang, J. S. Zhang, H. Z. Yu, Y. Wang, Y. T. Yan, J. L. Chen, J. Y. Zhao, Y. P. Zou

Abstract: We conducted observations of multiple HC3N (J = 10-9, 12-11, and 16-15) lines and the N2H+ (J = 1-0) line toward a large sample of 61 ultracompact (UC) H II regions, through the Institutde Radioastronomie Millmetrique 30 m and the Arizona Radio Observatory 12 m telescopes. The N2H+ J = 1-0 line is detected in 60 sources and HC3N is detected in 59 sources, including 40 sources with three lines, 9 s… ▽ More We conducted observations of multiple HC3N (J = 10-9, 12-11, and 16-15) lines and the N2H+ (J = 1-0) line toward a large sample of 61 ultracompact (UC) H II regions, through the Institutde Radioastronomie Millmetrique 30 m and the Arizona Radio Observatory 12 m telescopes. The N2H+ J = 1-0 line is detected in 60 sources and HC3N is detected in 59 sources, including 40 sources with three lines, 9 sources with two lines, and 10 sources with one line. Using the rotational diagram, the rotational temperature and column density of HC3N were estimated toward sources with at least two HC3N lines. For 10 sources with only one HC3N line, their parameters were estimated, taking one average value of Trot. For N2H+, we estimated the optical depth of the N2H+ J = 1-0 line, based on the line intensity ratio of its hyperfine structure lines. Then the excitation temperature and column density were calculated. When combining our results in UC H II regions and previous observation results on high-mass starless cores and high-mass protostellar cores, the N(HC3N)/N(N2H+) ratio clearly increases from the region stage. This means that the abundance ratio changes with the evolution of high-mass star-forming regions (HMSFRs). Moreover, positive correlations between the ratio and other evolutionary indicators (dust temperature, bolometric luminosity, and luminosity-to-mass ratio) are found. Thus we propose the ratio of N(HC3N)/N(N2H+) as a reliable chemical clock of HMSFRs. △ Less

Submitted 19 February, 2023; originally announced February 2023.

Comments: 40 pages, 8 figures and 8 tables

Journal ref: 2023ApJS..264...48W

Dynamic structure factor of one-dimensional Fermi superfluid with spin-orbit coupling

Authors: Zheng Gao, Lianyi He, Huaisong Zhao, Shi-Guo Peng, Peng Zou

Abstract: We theoretically calculate the density dynamic structure factor of one-dimensional Fermi superfluid with Raman-type spin-orbit coupling, and analyze its main dynamical character during phase transition between Bardeen-Cooper-Schrieffer superfluid and topological superfluid. Our theoretical results display four kinds of single-particle excitations induced by the two-branch structure of single-parti… ▽ More We theoretically calculate the density dynamic structure factor of one-dimensional Fermi superfluid with Raman-type spin-orbit coupling, and analyze its main dynamical character during phase transition between Bardeen-Cooper-Schrieffer superfluid and topological superfluid. Our theoretical results display four kinds of single-particle excitations induced by the two-branch structure of single-particle spectrum, and the cross single-particle excitation is much easier to be seen in the spin dynamic structure factor at a small transferred momentum. Also we find a new roton-like collective mode emerges at a fixed transferred momentum $q \simeq 2k_F$, and it only appears once the system enters the topological superfluid state. The occurrence of this roton-like excitation is related to switch of global minimum in single-particle spectrum from $k=0$ to $k \simeq 2k_F$. △ Less

Submitted 19 October, 2022; originally announced October 2022.

Journal ref: Phys. Rev. A 107, 013304 (2023)

Analyses of Flight Time During Solar Proton Events and Solar Flares

Authors: X. H. Xu, Y. Wang, F. S. Wei, X. S. Feng, M. H. Bo, H. W. Tang, D. S. Wang, B. Lei, B. Y. Wang, P. B. Zuo, C. W. Jiang, X. J. Xu, Z. L. Zhou, Z. Li, P. Zou, L. D. Wang, Y. X. Gu, Y. L. Chen, W. Y. Zhang, P. Sun

Abstract: Analyzing the effects of space weather on aviation is a new and develo** topic. It has been commonly accepted that the flight time of the polar flights may increase during solar proton events because the flights have to change their route to avoid the high-energy particles. However, apart from such phenomenon, researches related to the flight time during space weather events is very rare. Based… ▽ More Analyzing the effects of space weather on aviation is a new and develo** topic. It has been commonly accepted that the flight time of the polar flights may increase during solar proton events because the flights have to change their route to avoid the high-energy particles. However, apart from such phenomenon, researches related to the flight time during space weather events is very rare. Based on the analyses of 39 representative international air routes around westerlies, it is found that 97.44% (94.87%) of the commercial airplanes on the westbound (eastbound) air routes reveal shorter (longer) flight time during solar proton events compared to those during quiet periods, and the averaged magnitude of change in flight time is ~10 min or 0.21%-4.17% of the total flight durations. Comparative investigations reassure the certainty of such phenomenon that the directional differences in flight time are still incontrovertible regardless of over-land routes (China-Europe) or over-sea routes (China-Western America). Further analyses suggest that the solar proton events associated atmospheric heating will change the flight durations by weakening certain atmospheric circulations, such as the polar jet stream. While the polar jet stream will not be obviously altered during solar flares so that the directional differences in flight time are not found. Besides the conventional space weather effects already known, this paper is the first report that indicates a distinct new scenario of how the solar proton events affect flight time. These analyses are also important for aviation since our discoveries could help the airways optimize the air routes to save passenger time costs, reduce fuel costs and even contribute to the global warming issues. △ Less

Submitted 15 September, 2022; originally announced September 2022.

Comments: submitted to Scientific Reports

Characteristics of Flight Delays during Solar Flares

Authors: X. H. Xu, Y. Wang, F. S. Wei, X. S. Feng, M. H. Bo, H. W. Tang, D. S. Wang, L. Bian, B. Y. Wang, W. Y. Zhang, Y. S. Huang, Z. Li, J. P. Guo, P. B. Zuo, C. W. Jiang, X. J. Xu, Z. L. Zhou, P. Zou

Abstract: Solar flare is one of the severest solar activities on the sun, and it has many important impacts on the near-earth space. It has been found that flight arrival delays will increase during solar flare. However, the detailed intrinsic mechanism of how solar flares influence the delays is still unknown. Based on 5-years huge amount of flight data, here we comprehensively analyze the flight departure… ▽ More Solar flare is one of the severest solar activities on the sun, and it has many important impacts on the near-earth space. It has been found that flight arrival delays will increase during solar flare. However, the detailed intrinsic mechanism of how solar flares influence the delays is still unknown. Based on 5-years huge amount of flight data, here we comprehensively analyze the flight departure delays during 57 solar flares. It is found that the averaged flight departure delay time during solar flares increased by 20.68% (7.67 min) compared to those during quiet periods. It is also shown that solar flare related flight delays reveal apparent time and latitude dependencies. Flight delays during dayside solar flares are more serious than those during nightside flares, and the longer (shorter) delays tend to occur in the lower (higher) latitude airport. Further analyses suggest that flight delay time and delay rate would be directly modulated by the solar intensity (soft X-ray flux) and the Solar Zenith Angle. For the first time, these results indicate that the communication interferences caused by solar flares will directly affect flight departure delay time and delay rate. This work also expands our conventional understandings to the impacts of solar flares on human society, and it could also provide us with brand new views to help prevent or cope with flight delays. △ Less

Submitted 15 September, 2022; originally announced September 2022.

Comments: submitted to APJL

The Effects of Space Weather on Flight Delays

Authors: Y. Wang, X. H. Xu, F. S. Wei, X. S. Feng, M. H. Bo, H. W. Tang, D. S. Wang, L. Bian, B. Y. Wang, W. Y. Zhang, Y. S. Huang, Z. Li, J. P. Guo, P. B. Zuo, C. W. Jiang, X. J. Xu, Z. L. Zhou, P. Zou

Abstract: Although the sun is really far away from us, some solar activities could still influence the performance and reliability of space-borne and ground-based technological systems on Earth. Those time-varying conditions in space caused by the sun are also called space weather, as the atmospheric conditions that can affect weather on the ground. It is known that aviation activities can be affected durin… ▽ More Although the sun is really far away from us, some solar activities could still influence the performance and reliability of space-borne and ground-based technological systems on Earth. Those time-varying conditions in space caused by the sun are also called space weather, as the atmospheric conditions that can affect weather on the ground. It is known that aviation activities can be affected during space weather events, but the exact effects of space weather on aviation are still unclear. Especially how the flight delays, the top topic concerned by most people, will be affected by space weather has never been thoroughly researched. By analyzing huge amount of flight data (~5X106 records), for the first time, we demonstrate that space weather events could have systematically modulating effects on flight delays. The average arrival delay time and 30-minute delay rate during space weather events are significantly increased by 81.34% and 21.45% respectively compared to those during quiet periods. The evident negative correlation between the yearly flight regularity rate and the yearly mean total sunspot number during 22 years also confirms such delay effects. Further studies indicate that the interference in communication and navigation caused by geomagnetic field fluctuations and ionospheric disturbances associated with the space weather events will increase the flight delay time and delay rate. These results expand the traditional field of space weather research and could also provide us with brand new views for improving the flight delay predications. △ Less

Submitted 15 September, 2022; originally announced September 2022.

Comments: submitted to science advances

Spin-orbital-angular-momentum-coupled quantum gases

Authors: Shi-Guo Peng, Kaijun Jiang, Xiao-Long Chen, Ke-Ji Chen, Peng Zou, Lianyi He

Abstract: We briefly review the recent progress of theories and experiments on spin-orbital-angular-momentum (SOAM)-coupled quantum gases. The coupling between the intrinsic degree of freedom of particles and their external orbital motions widely exists in universe, and leads to a broad variety of fundamental phenomena both in the classical physics and quantum mechanics. Recent realization of synthetic SOAM… ▽ More We briefly review the recent progress of theories and experiments on spin-orbital-angular-momentum (SOAM)-coupled quantum gases. The coupling between the intrinsic degree of freedom of particles and their external orbital motions widely exists in universe, and leads to a broad variety of fundamental phenomena both in the classical physics and quantum mechanics. Recent realization of synthetic SOAM coupling in cold atoms has attracted a great deal of attention, and stimulates a large amount of considerations on exotic quantum phases in both Bose and Fermi gases. In this review, we present a basic idea of engineering SOAM coupling in neutral atoms, starting from a semiclassical description of atom-light interaction. Unique features of the single-particle physics in the presence of SOAM coupling are discussed. The intriguing ground-state quantum phases of weakly interacting Bose gases are introduced, with emphasis on a so-called angular stripe phase, which has yet been observed at present. It is demonstrated how to generate a stable giant vortex in a SOAM-coupled Fermi superfluid. We also discuss topological characters of a Fermi superfluid in the presence of SOAM coupling. We then introduce the experimental achievement of SOAM coupling in $^{87}$Rb Bose gases and its first observation of phase transitions. The most recent development of SOAM-coupled Bose gases in experiments is also summarized. Regarding the controllability of ultracold quantum gases, it opens a new era, on the quantum simulation point of view, to study the fundamental physics resulted from SOAM coupling as well as newly emergent quantum phases. △ Less

Submitted 5 November, 2022; v1 submitted 15 September, 2022; originally announced September 2022.

Comments: A brief review on the recent progress of spin-orbital-angular-momentum-coupled quantum gases. Comments are welcome

Journal ref: AAPPS Bulletin 32, 36 (2022)

Interstellar Nitrogen Isotope Ratios: New NH3 Data from the Galactic Center out to the Perseus Arm

Authors: J. L. Chen, J. S. Zhang, C. Henkel, Y. T. Yan, H. Z. Yu, J. J. Qiu, X. D. Tang, J. Wang, W. Liu, Y. X. Wang, Y. H. Zheng, J. Y. Zhao, Y. P. Zou

Abstract: Our aim is to measure the interstellar 14N/15N ratio across the Galaxy, to establish a standard data set on interstellar ammonia isotope ratios, and to provide new constraints on the Galactic chemical evolution. The (J, K ) = (1, 1), (2, 2), and (3, 3) lines of 14NH3 and 15NH3 were observed with the Shanghai Tianma 65 m radio telescope (TMRT) and the Effelsberg 100 m telescope toward a large sampl… ▽ More Our aim is to measure the interstellar 14N/15N ratio across the Galaxy, to establish a standard data set on interstellar ammonia isotope ratios, and to provide new constraints on the Galactic chemical evolution. The (J, K ) = (1, 1), (2, 2), and (3, 3) lines of 14NH3 and 15NH3 were observed with the Shanghai Tianma 65 m radio telescope (TMRT) and the Effelsberg 100 m telescope toward a large sample of 210 sources. One hundred fourty-one of these sources were detected by the TMRT in 14NH3. Eight of them were also detected in 15NH3. For 10 of the 36 sources with strong NH3 emission, the Effelsberg 100 m telescope successfully detected their 15NH3(1, 1) lines, including 3 sources (G081.7522, W51D, and Orion-KL) with detections by the TMRT telescope. Thus, a total of 15 sources are detected in both the 14NH3 and 15NH3 lines. Line and physical parameters for these 15 sources are derived, including optical depths, rotation and kinetic temperatures, and total column densities. 14N/15N isotope ratios were determined from the 14NH3/15NH3 abundance ratios. The isotope ratios obtained from both telescopes agree for a given source within the uncertainties, and no dependence on heliocentric distance and kinetic temperature is seen. 14N/15N ratios tend to increase with galactocentric distance, confirming a radial nitrogen isotope gradient. This is consistent with results from recent Galactic chemical model calculations, including the impact of superasymptotic giant branch stars and novae. △ Less

Submitted 8 August, 2022; originally announced August 2022.

Hybrid Multimodal Feature Extraction, Mining and Fusion for Sentiment Analysis

Authors: Jia Li, Ziyang Zhang, Junjie Lang, Yueqi Jiang, Liuwei An, Peng Zou, Yangyang Xu, Sheng Gao, Jie Lin, Chunxiao Fan, Xiao Sun, Meng Wang

Abstract: In this paper, we present our solutions for the Multimodal Sentiment Analysis Challenge (MuSe) 2022, which includes MuSe-Humor, MuSe-Reaction and MuSe-Stress Sub-challenges. The MuSe 2022 focuses on humor detection, emotional reactions and multimodal emotional stress utilizing different modalities and data sets. In our work, different kinds of multimodal features are extracted, including acoustic,… ▽ More In this paper, we present our solutions for the Multimodal Sentiment Analysis Challenge (MuSe) 2022, which includes MuSe-Humor, MuSe-Reaction and MuSe-Stress Sub-challenges. The MuSe 2022 focuses on humor detection, emotional reactions and multimodal emotional stress utilizing different modalities and data sets. In our work, different kinds of multimodal features are extracted, including acoustic, visual, text and biological features. These features are fused by TEMMA and GRU with self-attention mechanism frameworks. In this paper, 1) several new audio features, facial expression features and paragraph-level text embeddings are extracted for accuracy improvement. 2) we substantially improve the accuracy and reliability of multimodal sentiment prediction by mining and blending the multimodal features. 3) effective data augmentation strategies are applied in model training to alleviate the problem of sample imbalance and prevent the model from learning biased subject characters. For the MuSe-Humor sub-challenge, our model obtains the AUC score of 0.8932. For the MuSe-Reaction sub-challenge, the Pearson's Correlations Coefficient of our approach on the test set is 0.3879, which outperforms all other participants. For the MuSe-Stress sub-challenge, our approach outperforms the baseline in both arousal and valence on the test dataset, reaching a final combined result of 0.5151. △ Less

Submitted 12 August, 2022; v1 submitted 5 August, 2022; originally announced August 2022.

Comments: 8 pages, 2 figures, to appear in MuSe 2022 (ACM MM2022 co-located workshop)

Multi scale Feature Extraction and Fusion for Online Knowledge Distillation

Authors: Panpan Zou, Yinglei Teng, Tao Niu

Abstract: Online knowledge distillation conducts knowledge transfer among all student models to alleviate the reliance on pre-trained models. However, existing online methods rely heavily on the prediction distributions and neglect the further exploration of the representational knowledge. In this paper, we propose a novel Multi-scale Feature Extraction and Fusion method (MFEF) for online knowledge distilla… ▽ More Online knowledge distillation conducts knowledge transfer among all student models to alleviate the reliance on pre-trained models. However, existing online methods rely heavily on the prediction distributions and neglect the further exploration of the representational knowledge. In this paper, we propose a novel Multi-scale Feature Extraction and Fusion method (MFEF) for online knowledge distillation, which comprises three key components: Multi-scale Feature Extraction, Dual-attention and Feature Fusion, towards generating more informative feature maps for distillation. The multiscale feature extraction exploiting divide-and-concatenate in channel dimension is proposed to improve the multi-scale representation ability of feature maps. To obtain more accurate information, we design a dual-attention to strengthen the important channel and spatial regions adaptively. Moreover, we aggregate and fuse the former processed feature maps via feature fusion to assist the training of student models. Extensive experiments on CIF AR-10, CIF AR-100, and CINIC-10 show that MFEF transfers more beneficial representational knowledge for distillation and outperforms alternative methods among various network architectures △ Less

Submitted 16 June, 2022; originally announced June 2022.

Comments: 12 pages, 3 figures

Asymptotic Soft Cluster Pruning for Deep Neural Networks

Authors: Tao Niu, Yinglei Teng, Panpan Zou

Abstract: Filter pruning method introduces structural sparsity by removing selected filters and is thus particularly effective for reducing complexity. Previous works empirically prune networks from the point of view that filter with smaller norm contributes less to the final results. However, such criteria has been proven sensitive to the distribution of filters, and the accuracy may hard to recover since… ▽ More Filter pruning method introduces structural sparsity by removing selected filters and is thus particularly effective for reducing complexity. Previous works empirically prune networks from the point of view that filter with smaller norm contributes less to the final results. However, such criteria has been proven sensitive to the distribution of filters, and the accuracy may hard to recover since the capacity gap is fixed once pruned. In this paper, we propose a novel filter pruning method called Asymptotic Soft Cluster Pruning (ASCP), to identify the redundancy of network based on the similarity of filters. Each filter from over-parameterized network is first distinguished by clustering, and then reconstructed to manually introduce redundancy into it. Several guidelines of clustering are proposed to better preserve feature extraction ability. After reconstruction, filters are allowed to be updated to eliminate the effect caused by mistakenly selected. Besides, various decaying strategies of the pruning rate are adopted to stabilize the pruning process and improve the final performance as well. By gradually generating more identical filters within each cluster, ASCP can remove them through channel addition operation with almost no accuracy drop. Extensive experiments on CIFAR-10 and ImageNet datasets show that our method can achieve competitive results compared with many state-of-the-art algorithms. △ Less

Submitted 16 June, 2022; originally announced June 2022.

Cyanopolyyne line survey towards high-mass star-forming regions with TMRT

Authors: Y. X. Wang, J. S. Zhang, Y. T. Yan, J. J. Qiu, J. L. Chen, J. Y. Zhao, Y. P. Zou, X. C. Wu, X. L. He, Y. B. Gong, J. H. Cai

Abstract: We carried out a cyanopolyyne line survey towards a large sample of HMSFRs using the Shanghai Tian Ma 65m Radio Telescope (TMRT). Our sample consisted of 123 targets taken from the TMRT C band line survey. It included three kinds of sources, namely those with detection of the 6.7 GHz CH3OH maser alone, with detection of the radio recombination line (RRL) alone, and with detection of both (hereafte… ▽ More We carried out a cyanopolyyne line survey towards a large sample of HMSFRs using the Shanghai Tian Ma 65m Radio Telescope (TMRT). Our sample consisted of 123 targets taken from the TMRT C band line survey. It included three kinds of sources, namely those with detection of the 6.7 GHz CH3OH maser alone, with detection of the radio recombination line (RRL) alone, and with detection of both (hereafter referred to as Maser-only, RRL-only, and Maser-RRL sources, respectively). We detected HC3N in 38 sources, HC5N in 11 sources, and HC7N in G24.790+0.084, with the highest detection rate being found for Maser-RRL sources and a very low detection rate found for RRL-only sources. Their column densities were derived using the rotational temperature measured from the NH3 lines. And we constructed and fitted the far-infrared (FIR) spectral energy distributions. Based on these, we derive their dust temperatures, H2 column densities, and abundances of cyanopolyynes relative to H2. The detection rate, the column density, and the relative abundance of HC3N increase from Maser-only to Maser-RRL sources and decrease from Maser-RRL to RRL-only sources. This trend is consistent with the proposed evolutionary trend of HC3N under the assumption that our Maser-only, Maser-RRL, and RRL-only sources correspond to massive young stellar objects, ultra-compact HII regions, and normal classical HII regions, respectively. Furthermore, a statistical analysis of the integrated line intensity and column density of HC3N and shock-tracing molecules (SiO, H2CO) enabled us to find positive correlations between them. This suggests that HC3N may be another tracer of shocks, and should therefore be the subject of further observations and corresponding chemical simulations. Our results indirectly support the idea that the neutral--neutral reaction between C2H2 and CN is the dominant formation pathway of HC3N. △ Less

Submitted 16 May, 2022; originally announced May 2022.

Comments: 23 pages, 5 figures, 4 tables, Accepted to A&A

Journal ref: A&A 663, A177 (2022)

High Mixing Entropy Enhanced Hydrogen Evolution Reaction in AlMnYNiCoAu Catalysts

Authors: Peng Zou, Bowen Zang, Lijian Song, Juntao Huo, Jun-Qiang Wang

Abstract: An effective method for increasing the electrocatalytic activity of metallic glasses (MGs) in hydrogen evolution reaction (HER) is reported. This method applies a noble metal hybridization strategy to design a highly reactive catalyst for alkaline HER based on MGs with a nano-porous structure. The porous structure provides an abundance of active sites and exposes a large specific surface area to t… ▽ More An effective method for increasing the electrocatalytic activity of metallic glasses (MGs) in hydrogen evolution reaction (HER) is reported. This method applies a noble metal hybridization strategy to design a highly reactive catalyst for alkaline HER based on MGs with a nano-porous structure. The porous structure provides an abundance of active sites and exposes a large specific surface area to the electrolyte, thus effectively improving the activity of the HER catalysts. The do** of Au element can create electronic defects, adjust the electronic structure, change the electronic interaction, introduce lattice distortion, regulate mixing entropy, and improve the electrocatalytic performance of the catalyst. All these properties make np-AlMnYNiCoAu catalyst a potential electrode for alkaline HER. Significantly, we find that the high mixing entropy can enhance the HER performances. The present work could lead to a new approach to further develop environmentally friendly amorphous electrocatalyst with high efficiency and excellent stability for HER in alkaline solution. △ Less

Submitted 11 March, 2022; originally announced March 2022.

Probing two Higgs oscillations in a one-dimensional Fermi superfluid with Raman-type spin-orbit coupling

Authors: Genwang Fan, Xiao-Long Chen, Peng Zou

Abstract: We theoretically investigate the Higgs oscillation in a one-dimensional Raman-type spin-orbit-coupled Fermi superfluid with the time-dependent Bogoliubov-de Gennes equations. By linearly ram** or abruptly changing the effective Zeeman field in both the Bardeen-Cooper-Schrieffer state and the topological superfluid state, we find the amplitude of the order parameter exhibits an oscillating behavi… ▽ More We theoretically investigate the Higgs oscillation in a one-dimensional Raman-type spin-orbit-coupled Fermi superfluid with the time-dependent Bogoliubov-de Gennes equations. By linearly ram** or abruptly changing the effective Zeeman field in both the Bardeen-Cooper-Schrieffer state and the topological superfluid state, we find the amplitude of the order parameter exhibits an oscillating behaviour over time with two different frequencies (i.e., two Higgs oscillations) in contrast to the single one in a conventional Fermi superfluid. The observed period of oscillations has a great agreement with the one calculated using the previous prediction [Volkov and Kogan, J. Exp. Theor. Phys. 38, 1018 (1974)], where the oscillating periods are now determined by the minimums of two quasi-particle spectrum in this system. We further verify the existence of two Higgs oscillations using a periodic ramp strategy with theoretically calculated driving frequency. Our predictions would be useful for further theoretical and experimental studies of these Higgs oscillations in spin-orbit-coupled systems. △ Less

Submitted 6 March, 2022; originally announced March 2022.

Comments: 8 pages, 8 figures

Journal ref: Front. Phys. 17(5), 52502 (2022)

Community Trend Prediction on Heterogeneous Graph in E-commerce

Authors: Jiahao Yuan, Zhao Li, Pengcheng Zou, Xuan Gao, **wei Pan, Wendi Ji, Xiaoling Wang

Abstract: In online shop**, ever-changing fashion trends make merchants need to prepare more differentiated products to meet the diversified demands, and e-commerce platforms need to capture the market trend with a prophetic vision. For the trend prediction, the attribute tags, as the essential description of items, can genuinely reflect the decision basis of consumers. However, few existing works explore… ▽ More In online shop**, ever-changing fashion trends make merchants need to prepare more differentiated products to meet the diversified demands, and e-commerce platforms need to capture the market trend with a prophetic vision. For the trend prediction, the attribute tags, as the essential description of items, can genuinely reflect the decision basis of consumers. However, few existing works explore the attribute trend in the specific community for e-commerce. In this paper, we focus on the community trend prediction on the item attribute and propose a unified framework that combines the dynamic evolution of two graph patterns to predict the attribute trend in a specific community. Specifically, we first design a communityattribute bipartite graph at each time step to learn the collaboration of different communities. Next, we transform the bipartite graph into a hypergraph to exploit the associations of different attribute tags in one community. Lastly, we introduce a dynamic evolution component based on the recurrent neural networks to capture the fashion trend of attribute tags. Extensive experiments on three real-world datasets in a large e-commerce platform show the superiority of the proposed approach over several strong alternatives and demonstrate the ability to discover the community trend in advance. △ Less

Submitted 24 February, 2022; originally announced February 2022.

Comments: Published as a full paper at WSDM 2022

An Adaptive Device-Edge Co-Inference Framework Based on Soft Actor-Critic

Authors: Tao Niu, Yinglei Teng, Zhu Han, Panpan Zou

Abstract: Recently, the applications of deep neural network (DNN) have been very prominent in many fields such as computer vision (CV) and natural language processing (NLP) due to its superior feature extraction performance. However, the high-dimension parameter model and large-scale mathematical calculation restrict the execution efficiency, especially for Internet of Things (IoT) devices. Different from t… ▽ More Recently, the applications of deep neural network (DNN) have been very prominent in many fields such as computer vision (CV) and natural language processing (NLP) due to its superior feature extraction performance. However, the high-dimension parameter model and large-scale mathematical calculation restrict the execution efficiency, especially for Internet of Things (IoT) devices. Different from the previous cloud/edge-only pattern that brings huge pressure for uplink communication and device-only fashion that undertakes unaffordable calculation strength, we highlight the collaborative computation between the device and edge for DNN models, which can achieve a good balance between the communication load and execution accuracy. Specifically, a systematic on-demand co-inference framework is proposed to exploit the multi-branch structure, in which the pre-trained Alexnet is right-sized through \emph{early-exit} and partitioned at an intermediate DNN layer. The integer quantization is enforced to further compress transmission bits. As a result, we establish a new Deep Reinforcement Learning (DRL) optimizer-Soft Actor Critic for discrete (SAC-d), which generates the \emph{exit point}, \emph{partition point}, and \emph{compressing bits} by soft policy iterations. Based on the latency and accuracy aware reward design, such an optimizer can well adapt to the complex environment like dynamic wireless channel and arbitrary CPU processing, and is capable of supporting the 5G URLLC. Real-world experiment on Raspberry Pi 4 and PC shows the outperformance of the proposed solution. △ Less

Submitted 9 January, 2022; originally announced January 2022.

Beyond the Longest Letter-duplicated Subsequence Problem

Authors: Wenfeng Lai, Adiesha Liyanage, Binhai Zhu, Peng Zou

Abstract: Given a sequence $S$ of length $n$, a letter-duplicated subsequence is a subsequence of $S$ in the form of $x_1^{d_1}x_2^{d_2}\cdots x_k^{d_k}$ with $x_i\inΣ$, $x_j\neq x_{j+1}$ and $d_i\geq 2$ for all $i$ in $[k]$ and $j$ in $[k-1]$. A linear time algorithm for computing the longest letter-duplicated subsequence (LLDS) of $S$ can be easily obtained. In this paper, we focus on two variants of this… ▽ More Given a sequence $S$ of length $n$, a letter-duplicated subsequence is a subsequence of $S$ in the form of $x_1^{d_1}x_2^{d_2}\cdots x_k^{d_k}$ with $x_i\inΣ$, $x_j\neq x_{j+1}$ and $d_i\geq 2$ for all $i$ in $[k]$ and $j$ in $[k-1]$. A linear time algorithm for computing the longest letter-duplicated subsequence (LLDS) of $S$ can be easily obtained. In this paper, we focus on two variants of this problem. We first consider the constrained version when $Σ$ is unbounded, each letter appears in $S$ at least 6 times and all the letters in $Σ$ must appear in the solution. We show that the problem is NP-hard (a further twist indicates that the problem does not admit any polynomial time approximation). The reduction is from possibly the simplest version of SAT that is NP-complete, $(\leq 2,1,\leq 3)$-SAT, where each variable appears at most twice positively and exact once negatively, and each clause contains at most three literals and some clauses must contain exactly two literals. (We hope that this technique will serve as a general tool to help us proving the NP-hardness for some more tricky sequence problems involving only one sequence -- much harder than with at least two input sequences, which we apply successfully at the end of the paper on some extra variations of the LLDS problem.) We then show that when each letter appears in $S$ at most 3 times, then the problem admits a factor $1.5-O(\frac{1}{n})$ approximation. Finally, we consider the weighted version, where the weight of a block $x_i^{d_i} (d_i\geq 2)$ could be any positive function which might not grow with $d_i$. We give a non-trivial $O(n^2)$ time dynamic programming algorithm for this version, i.e., computing an LD-subsequence of $S$ whose weight is maximized. △ Less

Submitted 4 January, 2022; v1 submitted 10 December, 2021; originally announced December 2021.

Comments: 18 pages

MSC Class: 68W01; 68W32

MELONS: generating melody with long-term structure using transformers and structure graph

Authors: Yi Zou, Pei Zou, Yi Zhao, Kaixiang Zhang, Ran Zhang, Xiaorui Wang

Abstract: The creation of long melody sequences requires effective expression of coherent musical structure. However, there is no clear representation of musical structure. Recent works on music generation have suggested various approaches to deal with the structural information of music, but generating a full-song melody with clear long-term structure remains a challenge. In this paper, we propose MELONS,… ▽ More The creation of long melody sequences requires effective expression of coherent musical structure. However, there is no clear representation of musical structure. Recent works on music generation have suggested various approaches to deal with the structural information of music, but generating a full-song melody with clear long-term structure remains a challenge. In this paper, we propose MELONS, a melody generation framework based on a graph representation of music structure which consists of eight types of bar-level relations. MELONS adopts a multi-step generation method with transformer-based networks by factoring melody generation into two sub-problems: structure generation and structure conditional melody generation. Experimental results show that MELONS can produce structured melodies with high quality and rich contents. △ Less

Submitted 3 November, 2021; v1 submitted 11 October, 2021; originally announced October 2021.

Overage and Staleness Metrics for Status Update Systems

Authors: Peng Zou, ** Zhang, Xianglin Wei, Suresh Subramaniam

Abstract: Status update systems consist of sensors that take measurements of a physical parameter and transmit them to a remote receiver. Age of Information (AoI) has been studied extensively as a metric for the freshness of information in such systems with and without an enforced hard or soft deadline. In this paper, we propose three metrics for status update systems to measure the ability of different que… ▽ More Status update systems consist of sensors that take measurements of a physical parameter and transmit them to a remote receiver. Age of Information (AoI) has been studied extensively as a metric for the freshness of information in such systems with and without an enforced hard or soft deadline. In this paper, we propose three metrics for status update systems to measure the ability of different queuing systems to meet a threshold requirement for the AoI. The {\em overage probability} is defined as the probability that the age of the most recent update packet held by the receiver is larger than the threshold. The {\em stale update probability} is the probability that an update is stale, i.e., its age has exceeded the deadline, when it is delivered to the receiver. Finally, the {\em average overage} is defined as the time average of the overage (i.e., age beyond the threshold), and is a measure of the average ``staleness'' of the update packets held by the receiver. We investigate these metrics in three typical status update queuing systems -- M/G/1/1, M/G/1/$2^*$, and M/M/1. Numerical results show the performances for these metrics under different parameter settings and different service distributions. The differences between the average overage and average AoI are also shown. Our results demonstrate that a lower bound exists for the stale update probability when the buffer size is limited. Further, we observe that the overage probability decreases and the stale update probability increases as the update arrival rate increases. △ Less

Submitted 9 October, 2021; v1 submitted 28 September, 2021; originally announced September 2021.

Formation of Magnetic Flux Rope during Solar Eruption. I. Evolution of Toroidal Flux and Reconnection Flux

Authors: Chaowei Jiang, Jun Chen, Aiying Duan, Xinkai Bian, Xinyi Wang, Jiaying Li, Peng Zou, Xueshang Feng

Abstract: Magnetic flux ropes (MFRs) constitute the core structure of coronal mass ejections (CMEs), but hot debates remain on whether the MFR forms before or during solar eruptions. Furthermore, how flare reconnection shapes the erupting MFR is still elusive in three dimensions. Here we studied a new MHD simulation of CME initiation by tether-cutting magnetic reconnection in a single magnetic arcade. The s… ▽ More Magnetic flux ropes (MFRs) constitute the core structure of coronal mass ejections (CMEs), but hot debates remain on whether the MFR forms before or during solar eruptions. Furthermore, how flare reconnection shapes the erupting MFR is still elusive in three dimensions. Here we studied a new MHD simulation of CME initiation by tether-cutting magnetic reconnection in a single magnetic arcade. The simulation follows the whole life, including the birth and subsequent evolution, of an MFR during eruption. In the early phase, the MFR is partially separated from its ambient field by a magnetic quasi-separatrix layer (QSL) that has a double-J shaped footprint on the bottom surface. With the ongoing of the reconnection, the arms of the two J-shaped footprints continually separate from each other, and the hooks of the J shaped footprints expand and eventually become closed almost at the eruption peak time, and thereafter the MFR is fully separated from the un-reconnected field by the QSL. We further studied the evolution of the toroidal flux in the MFR and compared it with that of the reconnected flux. Our simulation reproduced an evolution pattern of increase-to-decrease of the toroidal flux, which is reported recently in observations of variations in flare ribbons and transient coronal dimming. The increase of toroidal flux is owing to the flare reconnection in the early phase that transforms the sheared arcade to twisted field lines, while its decrease is a result of reconnection between field lines in the interior of the MFR in the later phase. △ Less

Submitted 17 September, 2021; originally announced September 2021.

Comments: 10 pages, 5 figures, accepted by Frontiers in Physics

The Causes of Peripheral Coronal Loop Contraction and Disappearance Revealed in a Magnetohydrodynamic Simulation of Solar Eruption

Authors: Juntao Wang, Chaowei Jiang, Ding Yuan, Peng Zou

Abstract: The phenomenon of peripheral coronal loop contraction during solar flares and eruptions, recently discovered in observations, gradually intrigues solar physicists. However, its underlying physical mechanism is still uncertain. One is Hudson (2000)'s implosion conjecture which attributes it to magnetic pressure reduction in the magnetic energy liberation core, while other researchers proposed alter… ▽ More The phenomenon of peripheral coronal loop contraction during solar flares and eruptions, recently discovered in observations, gradually intrigues solar physicists. However, its underlying physical mechanism is still uncertain. One is Hudson (2000)'s implosion conjecture which attributes it to magnetic pressure reduction in the magnetic energy liberation core, while other researchers proposed alternative explanations. In previous observational studies we also note the disappearance of peripheral shrinking loops in the late phase, of which there is a lack of investigation and interpretation. In this paper, we exploit a full MHD simulation of solar eruption to study the causes of the two phenomena. It is found that the loop motion in the periphery is well correlated with magnetic energy accumulation and dissipation in the core, and the loop shrinkage is caused by a more significant reduction in magnetic pressure gradient force than in magnetic tension force, consistent with the implosion conjecture. The peripheral contracting loops in the late phase act as inflow to reconnect with central erupting structures, which destroys their identities and naturally explains their disappearance. We also propose a positive feedback between the peripheral magnetic reconnection and the central eruption. △ Less

Submitted 13 February, 2021; originally announced February 2021.

Comments: 13 pages, 8 figures, accept by ApJ

Dynamic structure factors of a strongly interacting Fermi superfluid near an orbital Feshbach resonance across the phase transition from BCS to Sarma superfluid

Authors: Peng Zou, Huaisong Zhao, Lianyi He, Xia-Ji Liu, Hui Hu

Abstract: We theoretically investigate dynamic structure factors of a strongly interacting Fermi superfluid near an orbital Feshbach resonance with random phase approximation, and find their dynamical characters during the phase transition between a balanced conventional Bardeen-Cooper-Schrieffer superfluid and a polarized Sarma superfluid by continuously varying the chemical potential difference of two spi… ▽ More We theoretically investigate dynamic structure factors of a strongly interacting Fermi superfluid near an orbital Feshbach resonance with random phase approximation, and find their dynamical characters during the phase transition between a balanced conventional Bardeen-Cooper-Schrieffer superfluid and a polarized Sarma superfluid by continuously varying the chemical potential difference of two spin components. In a BEC-like regime of the BCS superfluid, dynamic structure factors can do help to distinguish the in-phase ground state from the out-of-phase metastable state by the relative location of molecular excitation and Leggett mode, or the minimum energy to break a Cooper pair. In the phase transition between BCS and Sarma superfluid, we find the dynamic structure factor of Sarma superfluid has its own specific gapless excitation at a small transferred momentum which is mixed with the collective phonon excitation, and also a relatively strong atomic excitation at a large transferred momentum because of the existence of unpaired Fermi atoms, these signals can be used to differentiate Sarma superfluid from BCS superfluid. △ Less

Submitted 7 December, 2020; originally announced December 2020.

Comments: 9 pages, 7 figures

Journal ref: Phys. Rev. A 103, 053310 (2021)

Dynamical generation of solitons in one-dimensional Fermi superfluids with and without spin-orbit coupling

Authors: Lingchii Kong, Genwang Fan, Shi-Guo Peng, Xiao-Long Chen, Huaisong Zhao, Peng Zou

Abstract: We theoretically generalize a systematic language to describe the phase-imprinting technique to investigate the dynamical generation of solitons in a one-dimensional Raman-type spin-orbit-coupled Fermi superfluid. We check our method with the simulation of time-dependent Bogoliubov-de Gennes equations and find that our method not only can generate stable dark and even gray solitons in a convention… ▽ More We theoretically generalize a systematic language to describe the phase-imprinting technique to investigate the dynamical generation of solitons in a one-dimensional Raman-type spin-orbit-coupled Fermi superfluid. We check our method with the simulation of time-dependent Bogoliubov-de Gennes equations and find that our method not only can generate stable dark and even gray solitons in a conventional Fermi superfluid by controlling the transferred phase jump but also is feasible to create a stable dark soliton in both BCS and topological states of a spin-orbit-coupled Fermi superfluid. We also discuss the physical implication of our method. △ Less

Submitted 13 September, 2021; v1 submitted 28 November, 2020; originally announced November 2020.

Comments: 10 pages, 12 figures

Journal ref: Phys. Rev. A 103, 063318 (2021), Published 23 June 2021

Distant Supervision for E-commerce Query Segmentation via Attention Network

Authors: Zhao Li, Donghui Ding, Pengcheng Zou, Yu Gong, Xi Chen, Ji Zhang, Jianliang Gao, Youxi Wu, Yucong Duan

Abstract: The booming online e-commerce platforms demand highly accurate approaches to segment queries that carry the product requirements of consumers. Recent works have shown that the supervised methods, especially those based on deep learning, are attractive for achieving better performance on the problem of query segmentation. However, the lack of labeled data is still a big challenge for training a dee… ▽ More The booming online e-commerce platforms demand highly accurate approaches to segment queries that carry the product requirements of consumers. Recent works have shown that the supervised methods, especially those based on deep learning, are attractive for achieving better performance on the problem of query segmentation. However, the lack of labeled data is still a big challenge for training a deep segmentation network, and the problem of Out-of-Vocabulary (OOV) also adversely impacts the performance of query segmentation. Different from query segmentation task in an open domain, e-commerce scenario can provide external documents that are closely related to these queries. Thus, to deal with the two challenges, we employ the idea of distant supervision and design a novel method to find contexts in external documents and extract features from these contexts. In this work, we propose a BiLSTM-CRF based model with an attention module to encode external features, such that external contexts information, which can be utilized naturally and effectively to help query segmentation. Experiments on two datasets show the effectiveness of our approach compared with several kinds of baselines. △ Less

Submitted 8 November, 2020; originally announced November 2020.

Dynamical structure factors of a two-dimensional Fermi superfluid within random phase approximation

Authors: Huaisong Zhao, Xiaoxu Gao, Wen Liang, Peng Zou, Feng Yuan

Abstract: Based on random phase approximation (RPA), we numerically calculate dynamical structure factors of a balanced two-dimensional (2D) Fermi superfluid, and discuss their energy, momentum and interaction strength dependence in the 2D BEC-BCS crossover. At a small transferred momentum, a stable Higgs mode is observed in the unitary 2D Fermi superfluid gas where the particle-hole symmetry is not satisfi… ▽ More Based on random phase approximation (RPA), we numerically calculate dynamical structure factors of a balanced two-dimensional (2D) Fermi superfluid, and discuss their energy, momentum and interaction strength dependence in the 2D BEC-BCS crossover. At a small transferred momentum, a stable Higgs mode is observed in the unitary 2D Fermi superfluid gas where the particle-hole symmetry is not satisfied. Stronger interaction strength will make the visibility of the dispersion of Higgs mode harder to be observed. We also discuss the dimension effect and find that the signal of the Higgs mode in two dimension is more obvious than that in 3D case. At a large transferred momentum regime, stronger interaction strength will induce the weight of the molecules excitation increasing, while in verse the atomic one decreasing, which shows the pairing information of Fermi superfluid. The theoretical results qualitatively agree with the corresponding Quantum Monte Carlo data. △ Less

Submitted 31 July, 2020; originally announced July 2020.

Comments: 15 pages and 9 figures

Report number: https://doi.org/10.1088/1367-2630/abab3d

Journal ref: New Journal of Physics, 2020