Showing 1–7 of 7 results for author: Le, H H

I-MPN: Inductive Message Passing Network for Effective and Efficient Human-in-the-Loop Annotation of Mobile Eye Tracking Data

Authors: Hoang H. Le, Duy M. H. Nguyen, Omair Shahzad Bhatti, Laszlo Kopacsi, Thinh P. Ngo, Binh T. Nguyen, Michael Barz, Daniel Sonntag

Abstract: Understanding human visual processing in dynamic environments is essential for psychology and human-centered interaction design. Mobile eye-tracking systems, combining egocentric video and gaze signals, offer valuable insights. However, manual analysis of these recordings is time-intensive. In this work, we present a novel human-centered learning algorithm designed for automated object recognition… ▽ More Understanding human visual processing in dynamic environments is essential for psychology and human-centered interaction design. Mobile eye-tracking systems, combining egocentric video and gaze signals, offer valuable insights. However, manual analysis of these recordings is time-intensive. In this work, we present a novel human-centered learning algorithm designed for automated object recognition within mobile eye-tracking settings. Our approach seamlessly integrates an object detector with an inductive message-passing network technique (I-MPN), harnessing node features such as node profile information and positions. This integration enables our algorithm to learn embedding functions capable of generalizing to new object angle views, thereby facilitating rapid adaptation and efficient reasoning in dynamic contexts as users navigate through their environment. Through experiments conducted on three distinct video sequences, our \textit{interactive-based method} showcases significant performance improvements over fixed training/testing algorithms, even when trained on considerably smaller annotated samples collected through user feedback. Furthermore, we showcase exceptional efficiency in data annotation processes, surpassing approaches that use complete object detectors, combine detectors with convolutional networks, or employ interactive video segmentation. △ Less

Submitted 10 June, 2024; originally announced June 2024.

Comments: First version

Examining Monitoring System: Detecting Abnormal Behavior In Online Examinations

Authors: Dinh An Ngo, Thanh Dat Nguyen, Thi Le Chi Dang, Huy Hoan Le, Ton Bao Ho, Vo Thanh Khang Nguyen, Truong Thanh Hung Nguyen

Abstract: Cheating in online exams has become a prevalent issue over the past decade, especially during the COVID-19 pandemic. To address this issue of academic dishonesty, our "Exam Monitoring System: Detecting Abnormal Behavior in Online Examinations" is designed to assist proctors in identifying unusual student behavior. Our system demonstrates high accuracy and speed in detecting cheating in real-time s… ▽ More Cheating in online exams has become a prevalent issue over the past decade, especially during the COVID-19 pandemic. To address this issue of academic dishonesty, our "Exam Monitoring System: Detecting Abnormal Behavior in Online Examinations" is designed to assist proctors in identifying unusual student behavior. Our system demonstrates high accuracy and speed in detecting cheating in real-time scenarios, providing valuable information, and aiding proctors in decision-making. This article outlines our methodology and the effectiveness of our system in mitigating the widespread problem of cheating in online exams. △ Less

Submitted 19 February, 2024; originally announced February 2024.

Class-incremental Learning for Time Series: Benchmark and Evaluation

Authors: Zhongzheng Qiao, Quang Pham, Zhen Cao, Hoang H Le, P. N. Suganthan, Xudong Jiang, Ramasamy Savitha

Abstract: Real-world environments are inherently non-stationary, frequently introducing new classes over time. This is especially common in time series classification, such as the emergence of new disease classification in healthcare or the addition of new activities in human activity recognition. In such cases, a learning system is required to assimilate novel classes effectively while avoiding catastrophi… ▽ More Real-world environments are inherently non-stationary, frequently introducing new classes over time. This is especially common in time series classification, such as the emergence of new disease classification in healthcare or the addition of new activities in human activity recognition. In such cases, a learning system is required to assimilate novel classes effectively while avoiding catastrophic forgetting of the old ones, which gives rise to the Class-incremental Learning (CIL) problem. However, despite the encouraging progress in the image and language domains, CIL for time series data remains relatively understudied. Existing studies suffer from inconsistent experimental designs, necessitating a comprehensive evaluation and benchmarking of methods across a wide range of datasets. To this end, we first present an overview of the Time Series Class-incremental Learning (TSCIL) problem, highlight its unique challenges, and cover the advanced methodologies. Further, based on standardized settings, we develop a unified experimental framework that supports the rapid development of new algorithms, easy integration of new datasets, and standardization of the evaluation process. Using this framework, we conduct a comprehensive evaluation of various generic and time-series-specific CIL methods in both standard and privacy-sensitive scenarios. Our extensive experiments not only provide a standard baseline to support future research but also shed light on the impact of various design factors such as normalization layers or memory budget thresholds. Codes are available at https://github.com/zqiao11/TSCIL. △ Less

Submitted 19 February, 2024; originally announced February 2024.

Comments: Currently under review for KDD 2024 (ADS track)

Towards Effective Multi-Task Interaction for Entity-Relation Extraction: A Unified Framework with Selection Recurrent Network

Authors: An Wang, Ao Liu, Hieu Hanh Le, Haruo Yokota

Abstract: Entity-relation extraction aims to jointly solve named entity recognition (NER) and relation extraction (RE). Recent approaches use either one-way sequential information propagation in a pipeline manner or two-way implicit interaction with a shared encoder. However, they still suffer from poor information interaction due to the gap between the different task forms of NER and RE, raising a controve… ▽ More Entity-relation extraction aims to jointly solve named entity recognition (NER) and relation extraction (RE). Recent approaches use either one-way sequential information propagation in a pipeline manner or two-way implicit interaction with a shared encoder. However, they still suffer from poor information interaction due to the gap between the different task forms of NER and RE, raising a controversial question whether RE is really beneficial to NER. Motivated by this, we propose a novel and unified cascade framework that combines the advantages of both sequential information propagation and implicit interaction. Meanwhile, it eliminates the gap between the two tasks by reformulating entity-relation extraction as unified span-extraction tasks. Specifically, we propose a selection recurrent network as a shared encoder to encode task-specific independent and shared representations and design two sequential information propagation strategies to realize the sequential information flow between NER and RE. Extensive experiments demonstrate that our approaches can achieve state-of-the-art results on two common benchmarks, ACE05 and SciERC, and effectively model the multi-task interaction, which realizes significant mutual benefits of NER and RE. △ Less

Submitted 15 February, 2022; originally announced February 2022.

Protein-based microsphere biolasers fabricated by dehydration

Authors: Toan Van Nguyen, Nhat Van Pham, Hanh Hong Mai, Dung Chi Duong, Hai Hoang Le, Riccardo Sapienza, Van Duong Ta

Abstract: Biolasers made of biological materials have attracted a great of research attention due to their biocompatibility and biodegradability, which have the potential for biosensors and biointegration. However, the current fabrication method of biolasers suffers several limitations such as complicated processes, time-consuming and environmental unfriendly. In this work, a novel approach with green proce… ▽ More Biolasers made of biological materials have attracted a great of research attention due to their biocompatibility and biodegradability, which have the potential for biosensors and biointegration. However, the current fabrication method of biolasers suffers several limitations such as complicated processes, time-consuming and environmental unfriendly. In this work, a novel approach with green processes for fabricating solid-state microspheres biolasers is demonstrated. By using dehydration via a modified MicroglassificationTM technology, dye-doped bovine serum albumin (BSA) droplets can quickly (less than 10 minutes) and easily turn into solid microspheres with the diameter ranging from 10-150 μm. The size of the microspheres can be effectively controlled by changing either the concentration of BSA solution or the diameter of the initial droplets. Fabricated microspheres can act as efficient microlasers under optical pulse excitation. Lasing threshold of 7.8 μJ mm-2 and quality (Q) factor of about 1700 to 3100 are obtained. The size-dependence of lasing characteristics have been investigated and the results show a good agreement with whispering gallery mode (WGM) theory. Our finding contributes an effective technique for the fabrication of high Q factor microlasers that may be potential for applications in biological and chemical sensors. △ Less

Submitted 15 September, 2020; originally announced September 2020.

Journal ref: Soft Matter 15, 9721-9726 (2019)

Stochastic Variations in Nanoscale HZO based Ferroelectric finFETs: A Synergistic Approach of READ Optimization and Hybrid Precision Mixed Signal WRITE Operation to Mitigate the Implications on DNN Applications

Authors: Sourav De, Md. Aftab Baig, Bo-Han Qiu, Hoang- Hiep Le, Po-Jung Sung, Chun-Jung Su, Yao- Jen Lee, Darsen Lu

Abstract: This paper reports a synergistic approach of READ and WRITE optimization by deploying a high-precision digital computation unit along with a low-precision ferroelectric finFET (Fe-finFETs) based analog vector-matrix multiplication block for mitigating the impact of stochastic device variations in hafnium zirconium oxide (HZO) based Fe-finFETs. This paper reports a synergistic approach of READ and WRITE optimization by deploying a high-precision digital computation unit along with a low-precision ferroelectric finFET (Fe-finFETs) based analog vector-matrix multiplication block for mitigating the impact of stochastic device variations in hafnium zirconium oxide (HZO) based Fe-finFETs. △ Less

Submitted 8 May, 2021; v1 submitted 27 July, 2020; originally announced August 2020.

Vector relative degree and funnel control for differential-algebraic systems

Authors: Thomas Berger, Huy Hoàng Lê, Timo Reis

Abstract: We consider tracking control for multi-input multi-output differential-algebraic systems. First, the concept of vector relative degree is generalized for linear systems and we arrive at the novel concept of "truncated vector relative degree", and we derive a new normal form. Thereafter, we consider a class of nonlinear functional differential-algebraic systems which comprises linear systems with t… ▽ More We consider tracking control for multi-input multi-output differential-algebraic systems. First, the concept of vector relative degree is generalized for linear systems and we arrive at the novel concept of "truncated vector relative degree", and we derive a new normal form. Thereafter, we consider a class of nonlinear functional differential-algebraic systems which comprises linear systems with truncated vector relative degree. For this class we introduce a feedback controller which achieves that, for a given sufficiently smooth reference signal, the tracking error evolves within a prespecified performance funnel. We illustrate our results by an example of a robotic manipulator. △ Less

Submitted 15 January, 2020; originally announced January 2020.