Showing 1–11 of 11 results for author: Suo, X

Signed-Prompt: A New Approach to Prevent Prompt Injection Attacks Against LLM-Integrated Applications

Authors: Xuchen Suo

Abstract: The critical challenge of prompt injection attacks in Large Language Models (LLMs) integrated applications, a growing concern in the Artificial Intelligence (AI) field. Such attacks, which manipulate LLMs through natural language inputs, pose a significant threat to the security of these applications. Traditional defense strategies, including output and input filtering, as well as delimiter use, h… ▽ More The critical challenge of prompt injection attacks in Large Language Models (LLMs) integrated applications, a growing concern in the Artificial Intelligence (AI) field. Such attacks, which manipulate LLMs through natural language inputs, pose a significant threat to the security of these applications. Traditional defense strategies, including output and input filtering, as well as delimiter use, have proven inadequate. This paper introduces the 'Signed-Prompt' method as a novel solution. The study involves signing sensitive instructions within command segments by authorized users, enabling the LLM to discern trusted instruction sources. The paper presents a comprehensive analysis of prompt injection attack patterns, followed by a detailed explanation of the Signed-Prompt concept, including its basic architecture and implementation through both prompt engineering and fine-tuning of LLMs. Experiments demonstrate the effectiveness of the Signed-Prompt method, showing substantial resistance to various types of prompt injection attacks, thus validating its potential as a robust defense strategy in AI security. △ Less

Submitted 15 January, 2024; originally announced January 2024.

A new type of window functions constructed with exponential function

Authors: Haichao Xu, Xingpao Suo

Abstract: The Discrete Fourier Transform (DFT) is widely utilized for signal analysis but is plagued by spectral leakage, leading to inaccuracies in signal approximation. Window functions play a crucial role in mitigating spectral leakage by providing weighting mechanisms for discrete signals. In this paper, we introduce a novel window type based on exponential function, allowing for adjustable parameters a… ▽ More The Discrete Fourier Transform (DFT) is widely utilized for signal analysis but is plagued by spectral leakage, leading to inaccuracies in signal approximation. Window functions play a crucial role in mitigating spectral leakage by providing weighting mechanisms for discrete signals. In this paper, we introduce a novel window type based on exponential function, allowing for adjustable parameters and diverse variations. We present the formulation, properties, and motivation behind the design of the new window functions. Additionally, we analyze their behavior and evaluate their performance by comparing them with mainstream window functions using six parameters. Our findings demonstrate that these new window functions exhibit outstanding flexibility and versatility in signal analysis. △ Less

Submitted 23 December, 2023; originally announced December 2023.

Relieving the $S_8$ Tension: Exploring the Surface-type DBI Model as a Dark Matter Paradigm

Authors: Xingpao Suo, Xi Kang, Huanyuan Shan

Abstract: Recent observations from weak gravitational lensing (WL) surveys indicate a smoother Universe compared to the predictions of the Cosmic Microwave Background (CMB). This inconsistency is commonly referred to as the $σ_8$ tension or $S_8$ tension, where $σ_8$ represents the present root-mean-square matter fluctuation averaged over a sphere of radius $8 h^{-1} \mathrm{Mpc}$, and… ▽ More Recent observations from weak gravitational lensing (WL) surveys indicate a smoother Universe compared to the predictions of the Cosmic Microwave Background (CMB). This inconsistency is commonly referred to as the $σ_8$ tension or $S_8$ tension, where $σ_8$ represents the present root-mean-square matter fluctuation averaged over a sphere of radius $8 h^{-1} \mathrm{Mpc}$, and $S_8 \equiv σ_8\sqrt{Ω_m/0.3}$. In this article, we investigate a kind of general Dirac-Born-Infeld (DBI) Lagrangian referred to as \textit{surface-type DBI} (sDBI) model. We find that up to the linear order, the constraints on the sDBI model with high-redshift probe (CMB) and low-redshift probes (WL and Galaxy Clustering, GC) yield $S_8= 0.7448_{-0.21}^{+0.031}$ and $0.7426_{-0.085}^{+0.054}$, respectively. Remarkably, these values not only demonstrate self-consistency but also align with the values obtained from the majority of low-redshift probes. Furthermore, we present a discussion on exploring the non-linear effects of this model, which holds the potential to address additional challenges associated with Cold Dark Matter (CDM) on small scales. △ Less

Submitted 18 November, 2023; v1 submitted 10 July, 2023; originally announced July 2023.

Comments: Accepted by Physical Review D

The spherical Fast Multipole Method (sFMM) for Gravitational Lensing Simulation

Authors: Xingpao Suo, Xi Kang, Chengliang Wei, Guoliang Li

Abstract: In this paper, we present a spherical Fast Multipole Method (sFMM) for ray tracing simulation of gravitational lensing (GL) on a curved sky. The sFMM is a non-trivial extension of the Fast Multiple Method (FMM) to sphere $\mathbb S^2$, and it can accurately solve the Poisson equation with time complexity of $O(N)\log(N)$, where $N$ is the number of particles. It is found that the time complexity o… ▽ More In this paper, we present a spherical Fast Multipole Method (sFMM) for ray tracing simulation of gravitational lensing (GL) on a curved sky. The sFMM is a non-trivial extension of the Fast Multiple Method (FMM) to sphere $\mathbb S^2$, and it can accurately solve the Poisson equation with time complexity of $O(N)\log(N)$, where $N$ is the number of particles. It is found that the time complexity of the sFMM is near $O(N)$ and the computational accuracy can reach $10^{-10}$ in our test. In addition, compared with the Fast Spherical Harmonic Transform (FSHT), the sFMM is not only faster but more accurate, as it has the ability to reserve high-frequency components of the density field. These merits make the sFMM an optimum method to simulate the gravitational lensing on a curved sky, which is the case for upcoming large-area sky surveys, such as the Vera Rubin Observatory and the China Space Station Telescope. △ Less

Submitted 5 March, 2023; v1 submitted 13 October, 2022; originally announced October 2022.

Comments: Accepted by The Astrophysical Journal (ApJ)

A minimalistic approach to simulate multiple failure mechanisms of metal matrix fiber-reinforced composites by the phase field and cohesive zone models

Authors: Zhaoyang Hu, Xufei Suo, Feng Jiang, Yongxing Shen

Abstract: The mechanical properties of metal matrix fiber-reinforced composites depend on many aspects of their structure in a complicated way. In this paper, we propose a \emph{minimalistic} approach to study interface debonding, matrix cracking, and their competition in metal matrix fiber-reinforced elastoplastic composites by numerical simulation. This approach combines a cohesive zone model for interfac… ▽ More The mechanical properties of metal matrix fiber-reinforced composites depend on many aspects of their structure in a complicated way. In this paper, we propose a \emph{minimalistic} approach to study interface debonding, matrix cracking, and their competition in metal matrix fiber-reinforced elastoplastic composites by numerical simulation. This approach combines a cohesive zone model for interface debonding and a phase field model for matrix cracking. The features of this framework are: (1) crack nucleation, propagation, and branching can be easily tracked without the need of geometric programming; (2) the interface debonding is determined merely by the CZM, but not interfered by the phase field in the bulk; (3) the cohesive interface has zero thickness instead of being regularized; (4) any reasonable cohesive law of interest is readily incorporated with very few constraints; (5) elastoplasticity of the matrix is conveniently taken into account, as strains in the model are all well defined; (6) the competition of the two failure mechanisms, namely, matrix cracking and interface debonding, is accurately captured. Accuracy of this framework is verified with existing analytical and numerical results. The proposed framework shows a potential in investigating various complicated crack behaviors in composites. △ Less

Submitted 19 October, 2021; originally announced October 2021.

iButter: Neural Interactive Bullet Time Generator for Human Free-viewpoint Rendering

Authors: Liao Wang, Ziyu Wang, Pei Lin, Yuheng Jiang, Xin Suo, Minye Wu, Lan Xu, **gyi Yu

Abstract: Generating ``bullet-time'' effects of human free-viewpoint videos is critical for immersive visual effects and VR/AR experience. Recent neural advances still lack the controllable and interactive bullet-time design ability for human free-viewpoint rendering, especially under the real-time, dynamic and general setting for our trajectory-aware task. To fill this gap, in this paper we propose a neura… ▽ More Generating ``bullet-time'' effects of human free-viewpoint videos is critical for immersive visual effects and VR/AR experience. Recent neural advances still lack the controllable and interactive bullet-time design ability for human free-viewpoint rendering, especially under the real-time, dynamic and general setting for our trajectory-aware task. To fill this gap, in this paper we propose a neural interactive bullet-time generator (iButter) for photo-realistic human free-viewpoint rendering from dense RGB streams, which enables flexible and interactive design for human bullet-time visual effects. Our iButter approach consists of a real-time preview and design stage as well as a trajectory-aware refinement stage. During preview, we propose an interactive bullet-time design approach by extending the NeRF rendering to a real-time and dynamic setting and getting rid of the tedious per-scene training. To this end, our bullet-time design stage utilizes a hybrid training set, light-weight network design and an efficient silhouette-based sampling strategy. During refinement, we introduce an efficient trajectory-aware scheme within 20 minutes, which jointly encodes the spatial, temporal consistency and semantic cues along the designed trajectory, achieving photo-realistic bullet-time viewing experience of human activities. Extensive experiments demonstrate the effectiveness of our approach for convenient interactive bullet-time design and photo-realistic human free-viewpoint video generation. △ Less

Submitted 12 August, 2021; originally announced August 2021.

Comments: Accepted by ACM MM 2021

NeuralHumanFVV: Real-Time Neural Volumetric Human Performance Rendering using RGB Cameras

Authors: Xin Suo, Yuheng Jiang, Pei Lin, Yingliang Zhang, Kaiwen Guo, Minye Wu, Lan Xu

Abstract: 4D reconstruction and rendering of human activities is critical for immersive VR/AR experience.Recent advances still fail to recover fine geometry and texture results with the level of detail present in the input images from sparse multi-view RGB cameras. In this paper, we propose NeuralHumanFVV, a real-time neural human performance capture and rendering system to generate both high-quality geomet… ▽ More 4D reconstruction and rendering of human activities is critical for immersive VR/AR experience.Recent advances still fail to recover fine geometry and texture results with the level of detail present in the input images from sparse multi-view RGB cameras. In this paper, we propose NeuralHumanFVV, a real-time neural human performance capture and rendering system to generate both high-quality geometry and photo-realistic texture of human activities in arbitrary novel views. We propose a neural geometry generation scheme with a hierarchical sampling strategy for real-time implicit geometry inference, as well as a novel neural blending scheme to generate high resolution (e.g., 1k) and photo-realistic texture results in the novel views. Furthermore, we adopt neural normal blending to enhance geometry details and formulate our neural geometry and texture rendering into a multi-task learning framework. Extensive experiments demonstrate the effectiveness of our approach to achieve high-quality geometry and photo-realistic free view-point reconstruction for challenging human performances. △ Less

Submitted 13 March, 2021; originally announced March 2021.

Unique weak solutions of the non-resistive magnetohydrodynamic equations with fractional dissipation

Authors: Quansen Jiu, Xiaoxiao Suo, Jiahong Wu, Huan Yu

Abstract: This paper examines the uniqueness of weak solutions to the d-dimensional magnetohydrodynamic (MHD) equations with the fractional dissipation $(-Δ)^αu$ and without the magnetic diffusion. Important progress has been made on the standard Laplacian dissipation case $α=1$. This paper discovers that there are new phenomena with the case $α<1$. The approach for $α=1$ can not be directly extended to… ▽ More This paper examines the uniqueness of weak solutions to the d-dimensional magnetohydrodynamic (MHD) equations with the fractional dissipation $(-Δ)^αu$ and without the magnetic diffusion. Important progress has been made on the standard Laplacian dissipation case $α=1$. This paper discovers that there are new phenomena with the case $α<1$. The approach for $α=1$ can not be directly extended to $α<1$. We establish that, for $α<1$, any initial data $(u_0, b_0)$ in the inhomogeneous Besov space $B^σ_{2,\infty}(\mathbb R^d)$ with $σ> 1+\frac{d}{2}-α$ leads to a unique local solution. For the case $α\ge 1$, $u_0$ in the homogeneous Besov space $\mathring B^{1+\frac{d}{2}-2α}_{2,1}(\mathbb R^d)$ and $b_0$ in $ \mathring B^{1+\frac{d}{2}-α}_{2,1}(\mathbb R^d)$ guarantees the existence and uniqueness. These regularity requirements appear to be optimal. △ Less

Submitted 11 April, 2019; originally announced April 2019.

Comments: 35 pages

Sparse canonical correlation analysis

Authors: Xiaotong Suo, Victor Minden, Bradley Nelson, Robert Tibshirani, Michael Saunders

Abstract: Canonical correlation analysis was proposed by Hotelling [6] and it measures linear relationship between two multidimensional variables. In high dimensional setting, the classical canonical correlation analysis breaks down. We propose a sparse canonical correlation analysis by adding l1 constraints on the canonical vectors and show how to solve it efficiently using linearized alternating direction… ▽ More Canonical correlation analysis was proposed by Hotelling [6] and it measures linear relationship between two multidimensional variables. In high dimensional setting, the classical canonical correlation analysis breaks down. We propose a sparse canonical correlation analysis by adding l1 constraints on the canonical vectors and show how to solve it efficiently using linearized alternating direction method of multipliers (ADMM) and using TFOCS as a black box. We illustrate this idea on simulated data. △ Less

Submitted 2 June, 2017; v1 submitted 30 May, 2017; originally announced May 2017.

An Ordered Lasso and Sparse Time-Lagged Regression

Authors: Xiaotong Suo, Robert Tibshirani

Abstract: We consider regression scenarios where it is natural to impose an order constraint on the coefficients. We propose an order-constrained version of L1-regularized regression for this problem, and show how to solve it efficiently using the well-known Pool Adjacent Violators Algorithm as its proximal operator. The main application of this idea is time-lagged regression, where we predict an outcome at… ▽ More We consider regression scenarios where it is natural to impose an order constraint on the coefficients. We propose an order-constrained version of L1-regularized regression for this problem, and show how to solve it efficiently using the well-known Pool Adjacent Violators Algorithm as its proximal operator. The main application of this idea is time-lagged regression, where we predict an outcome at time t from features at the previous K time points. In this setting it is natural to assume that the coefficients decay as we move farther away from t, and hence the order constraint is reasonable. Potential applications include financial time series and prediction of dynamic patient out- comes based on clinical measurements. We illustrate this idea on real and simulated data. △ Less

Submitted 3 June, 2014; v1 submitted 25 May, 2014; originally announced May 2014.

Comments: 15 pages, 6 figures

Preparation and characterization of magnesium coating deposited by cold spraying

Authors: Xinkun Suo, Xue** Guo, Wenya Li, Marie-Pierre Planche, Rodolphe Bolot, Hanlin Liao, Christian Codde

Abstract: Magnesium (Mg) and its alloys have a great potential as structural materials due to their beneficial combination of high strength to weight ratio, high thermal conductivity and good machinability. However, few literatures about Mg coatings fabricated by cold spraying can be found. In this study, Mg coatings were fabricated by cold spraying, and the microstructure, phase structure, oxygen content a… ▽ More Magnesium (Mg) and its alloys have a great potential as structural materials due to their beneficial combination of high strength to weight ratio, high thermal conductivity and good machinability. However, few literatures about Mg coatings fabricated by cold spraying can be found. In this study, Mg coatings were fabricated by cold spraying, and the microstructure, phase structure, oxygen content and microhardness of the coating prepared under different main gas temperatures were investigated. The critical velocity of the particle was evaluated through numerical simulations. The particle deformation behavior and bonding mechanism were discussed. The result of the oxygen content test shows that the oxygen contents of the coatings did not increase comparing with that of the feedstock powder. The simulation results show that the critical velocity of Mg particles was in a range of 653 m/s to 677 m/s. The observation of the coating fracture morphology shows that the formation of the coating was due to the intensive plastic deformation and mechanical interlocking. The microhardness of the coating increased with the increase of the main gas temperature from 350oC to 450oC due to the decrease of the coating porosity. △ Less

Submitted 20 January, 2011; originally announced January 2011.

Comments: 20 pages, 9 figurs