arXiv subjects

Ang Li

Publications and source records attributed to Ang Li.

At least 289 records · Page 16Linked to original sources

arXiv2024

Rethinking Pruning for Vision-Language Models: Strategies for Effective Sparsity and Performance Restoration

Vision-Language Models (VLMs) integrate information from multiple modalities and have shown remarkable success across various tasks. However, deploying large-scale VLMs in resource-constrained scenarios is challenging. Pruning followed by finetuning offers a potential solution but remains underexplored for VLMs. This study addresses two key questions: how to distribute sparsity across different modality-specific models, and how to restore the performance of pruned sparse VLMs. Our preliminary studies identified two effective pruning settings: applying the same sparsity to both vision and language models, and pruning only the language models. While LoRA finetuning aims to restore sparse models, it faces challenges due to incompatibility with sparse models, disrupting the pruned sparsity. To overcome these issues, we propose SparseLoRA, which applies sparsity directly to LoRA weights. Our experimental results demonstrate significant improvements, including an 11.3\% boost under 2:4 sparsity and a 47.6\% enhancement under unstructured 70\% sparsity. Code is released at: \url{https://github.com/Shwai-He/VLM-Compression}.

cs.LG↗

arXiv2024

PID: Prompt-Independent Data Protection Against Latent Diffusion Models

The few-shot fine-tuning of Latent Diffusion Models (LDMs) has enabled them to grasp new concepts from a limited number of images. However, given the vast amount of personal images accessible online, this capability raises critical concerns about civil privacy. While several previous defense methods have been developed to prevent such misuse of LDMs, they typically assume that the textual prompts used by data protectors exactly match those employed by data exploiters. In this paper, we first empirically demonstrate that breaking this assumption, i.e., in cases where discrepancies exist between the textual conditions used by protectors and exploiters, could substantially reduce the effectiveness of these defenses. Furthermore, considering the visual encoder's independence from textual prompts, we delve into the visual encoder and thoroughly investigate how manipulating the visual encoder affects the few-shot fine-tuning process of LDMs. Drawing on these insights, we propose a simple yet effective method called \textbf{Prompt-Independent Defense (PID)} to safeguard privacy against LDMs. We show that PID can act as a strong privacy shield on its own while requiring significantly less computational power. We believe our studies, along with the comprehensive understanding and new defense method, provide a notable advance toward reliable data protection against LDMs.

cs.CR↗

arXiv2024

Multi-modal Stance Detection: New Datasets and Model

Stance detection is a challenging task that aims to identify public opinion from social media platforms with respect to specific targets. Previous work on stance detection largely focused on pure texts. In this paper, we study multi-modal stance detection for tweets consisting of texts and images, which are prevalent in today's fast-growing social media platforms where people often post multi-modal messages. To this end, we create five new multi-modal stance detection datasets of different domains based on Twitter, in which each example consists of a text and an image. In addition, we propose a simple yet effective Targeted Multi-modal Prompt Tuning framework (TMPT), where target information is leveraged to learn multi-modal stance features from textual and visual modalities. Experimental results on our five benchmark datasets show that the proposed TMPT achieves state-of-the-art performance in multi-modal stance detection.

cs.CL↗

arXiv2024

A New Solution for MU-MISO Symbol-Level Precoding: Extrapolation and Deep Unfolding

Constructive interference (CI) precoding, which converts the harmful multi-user interference into beneficial signals, is a promising and efficient interference management scheme in multi-antenna communication systems. However, CI-based symbol-level precoding (SLP) experiences high computational complexity as the number of symbol slots increases within a transmission block, rendering it unaffordable in practical communication systems. In this paper, we propose a symbol-level extrapolation (SLE) strategy to extrapolate the precoding matrix by leveraging the relationship between different symbol slots within in a transmission block, during which the channel state information (CSI) remains constant, where we design a closed-form iterative algorithm based on SLE for both PSK and QAM modulation. In order to further reduce the computational complexity, a sub-optimal closed-form solution based on SLE is further developed for PSK and QAM, respectively. Moreover, we design an unsupervised SLE-based neural network (SLE-Net) to unfold the proposed iterative algorithm, which helps enhance the interpretability of the neural network. By carefully designing the loss function of the SLE-Net, the time-complexity of the network can be reduced effectively. Extensive simulation results illustrate that the proposed algorithms can dramatically reduce the computational complexity and time complexity with only marginal performance loss, compared with the conventional SLP design methods.

eess.SP↗

arXiv2024

Jailbreak and Guard Aligned Language Models with Only Few In-Context Demonstrations

Large Language Models (LLMs) have shown remarkable success in various tasks, yet their safety and the risk of generating harmful content remain pressing concerns. In this paper, we delve into the potential of In-Context Learning (ICL) to modulate the alignment of LLMs. Specifically, we propose the In-Context Attack (ICA) which employs harmful demonstrations to subvert LLMs, and the In-Context Defense (ICD) which bolsters model resilience through examples that demonstrate refusal to produce harmful responses. We offer theoretical insights to elucidate how a limited set of in-context demonstrations can pivotally influence the safety alignment of LLMs. Through extensive experiments, we demonstrate the efficacy of ICA and ICD in respectively elevating and mitigating the success rates of jailbreaking prompts. Our findings illuminate the profound influence of ICL on LLM behavior, opening new avenues for improving the safety of LLMs.

cs.LG↗

arXiv2024

SiDA-MoE: Sparsity-Inspired Data-Aware Serving for Efficient and Scalable Large Mixture-of-Experts Models

Mixture-of-Experts (MoE) has emerged as a favorable architecture in the era of large models due to its inherent advantage, i.e., enlarging model capacity without incurring notable computational overhead. Yet, the realization of such benefits often results in ineffective GPU memory utilization, as large portions of the model parameters remain dormant during inference. Moreover, the memory demands of large models consistently outpace the memory capacity of contemporary GPUs. Addressing this, we introduce SiDA-MoE ($\textbf{S}$parsity-$\textbf{i}$nspired $\textbf{D}$ata-$\textbf{A}$ware), an efficient inference approach tailored for large MoE models. SiDA-MoE judiciously exploits both the system's main memory, which is now abundant and readily scalable, and GPU memory by capitalizing on the inherent sparsity on expert activation in MoE models. By adopting a data-aware perspective, SiDA-MoE achieves enhanced model efficiency with a neglectable performance drop. Specifically, SiDA-MoE attains a remarkable speedup in MoE inference with up to $3.93\times$ throughput increasing, up to $72\%$ latency reduction, and up to $80\%$ GPU memory saving with down to $1\%$ performance drop. This work paves the way for scalable and efficient deployment of large MoE models, even with constrained resources. Code is available at: https://github.com/timlee0212/SiDA-MoE.

cs.LG↗

arXiv2024

Benchmarking Optimizers for Qumode State Preparation with Variational Quantum Algorithms

Quantum state preparation involves preparing a target state from an initial system, a process integral to applications such as quantum machine learning and solving systems of linear equations. Recently, there has been a growing interest in qumodes due to advancements in the field and their potential applications. However there is a notable gap in the literature specifically addressing this area. This paper aims to bridge this gap by providing performance benchmarks of various optimizers used in state preparation with Variational Quantum Algorithms. We conducted extensive testing across multiple scenarios, including different target states, both ideal and sampling simulations, and varying numbers of basis gate layers. Our evaluations offer insights into the complexity of learning each type of target state and demonstrate that some optimizers perform better than others in this context. Notably, the Powell optimizer was found to be exceptionally robust against sampling errors, making it a preferred choice in scenarios prone to such inaccuracies. Additionally, the Simultaneous Perturbation Stochastic Approximation optimizer was distinguished for its efficiency and ability to handle increased parameter dimensionality effectively.

quant-ph↗

arXiv2024

Scalable Circuit Cutting and Scheduling in a Resource-constrained and Distributed Quantum System

Despite quantum computing's rapid development, current systems remain limited in practical applications due to their limited qubit count and quality. Various technologies, such as superconducting, trapped ions, and neutral atom quantum computing technologies are progressing towards a fault tolerant era, however they all face a diverse set of challenges in scalability and control. Recent efforts have focused on multi-node quantum systems that connect multiple smaller quantum devices to execute larger circuits. Future demonstrations hope to use quantum channels to couple systems, however current demonstrations can leverage classical communication with circuit cutting techniques. This involves cutting large circuits into smaller subcircuits and reconstructing them post-execution. However, existing cutting methods are hindered by lengthy search times as the number of qubits and gates increases. Additionally, they often fail to effectively utilize the resources of various worker configurations in a multi-node system. To address these challenges, we introduce FitCut, a novel approach that transforms quantum circuits into weighted graphs and utilizes a community-based, bottom-up approach to cut circuits according to resource constraints, e.g., qubit counts, on each worker. FitCut also includes a scheduling algorithm that optimizes resource utilization across workers. Implemented with Qiskit and evaluated extensively, FitCut significantly outperforms the Qiskit Circuit Knitting Toolbox, reducing time costs by factors ranging from 3 to 2000 and improving resource utilization rates by up to 3.88 times on the worker side, achieving a system-wide improvement of 2.86 times.

quant-ph↗

arXiv2024

Adversarial Examples Are Not Real Features

The existence of adversarial examples has been a mystery for years and attracted much interest. A well-known theory by \citet{ilyas2019adversarial} explains adversarial vulnerability from a data perspective by showing that one can extract non-robust features from adversarial examples and these features alone are useful for classification. However, the explanation remains quite counter-intuitive since non-robust features are mostly noise features to humans. In this paper, we re-examine the theory from a larger context by incorporating multiple learning paradigms. Notably, we find that contrary to their good usefulness under supervised learning, non-robust features attain poor usefulness when transferred to other self-supervised learning paradigms, such as contrastive learning, masked image modeling, and diffusion models. It reveals that non-robust features are not really as useful as robust or natural features that enjoy good transferability between these paradigms. Meanwhile, for robustness, we also show that naturally trained encoders from robust features are largely non-robust under AutoAttack. Our cross-paradigm examination suggests that the non-robust features are not really useful but more like paradigm-wise shortcuts, and robust features alone might be insufficient to attain reliable model robustness. Code is available at \url{https://github.com/PKU-ML/AdvNotRealFeatures}.

cs.LG↗

arXiv2024

Thermal x-ray studies of neutron stars and the equation of state

The understanding of neutron star equation of state hinges on a comprehensive analysis of multi-messenger, multi-wavelength data. The recent scrutiny of PSR J0030+0451 data by NICER introduces complexities, unveiling a tension with another X-ray observation of the central compact object in HESS J1731-347, specifically concerning the mass-radius constraint of low-mass neutron stars. This tension persists when integrating NICER's updated data with LIGO/Virgo's gravitational-wave data from the GW170817 binary neutron star merger. Despite attempts to reconcile these disparate observations, the current combined data still can not distinguish different types of neutron stars -- whether they are pure neutron stars or hybrid stars. Bayesian inference indicates only modest changes in the posterior ranges of parameters related to the nuclear matter and deconfinement phase transition. This ongoing exploration underscores the intricate challenges in precisely characterizing neutron stars. It also points out that it is possible to probe the equation of state at different density regimes from future more accurate radii of neutron stars with various masses.

astro-ph.HE↗

arXiv2024

Design of an entanglement purification protocol selection module

Entanglement purification protocols, designed to improve the fidelity of Bell states over quantum networks for inter-node communications, have attracted significant attention over the last few decades. These protocols have great potential to resolve a core challenge in quantum networking of generating high-fidelity Bell states. However, previous studies focused on the theoretical discussion with limited consideration of realistic errors. Studies of dynamically selecting the right purification protocol under various realistic errors that populate in practice have yet to be performed. In this work, we study the performance of various purification protocols under realistic errors by conducting density matrix simulations over a large suite of error models. Based on our findings of how specific error channels affect the performance of purification protocols, we propose a module that can be embedded in the quantum network. This module determines and selects the appropriate purification protocol, considering not only expected specifications from the network layer but also the capabilities of the physical layer. Finally, the performance of our proposed module is verified using two benchmark categories. Compared with the default approach and exhaustive search approach, we show a success rate approaching 90% in identifying the optimal purification protocol for our target applications.

quant-ph↗

arXiv2024

Numerical study of transitions in lid-driven flows in semicircular cavities

In this article, three-dimensional (3D) lid-driven flows in semicircular cavities are studied. The numerical solution of the Navier-Stokes equations modeling incompressible viscous fluid flow in cavities is obtained via a methodology combining a first-order accurate operator-splitting scheme, a fictitious domain formulation, and finite element space approximations. The critical Reynolds numbers (Re_{cr}) for having oscillatory flow (a Hopf bifurcation) are obtained. The associated oscillating motion in a semicircular cavity with length equal to width has been studied in detail. Based on the averaged velocity field in one period of oscillating motion, the flow difference (called oscillation mode) between the velocity field and averaged one at several time instances in such period shows almost the same flow pattern for the Reynolds numbers close to Re_{cr}. This oscillation mode in a semicircular cavity shows a close similarity to the one obtained in a shallow cavity, but with some difference in a shallow cavity which is triggered by the presence of two vertical side walls and downstream wall.

physics.flu-dyn↗

arXiv2024

AQM: A Refresh of the Abstract Qubit Model for Quantum Computing Co-design

Qubits are the fundamental building blocks of quantum information science and applications, whose concept is widely utilized in both quantum physics and quantum computation. While the significance of qubits and their implementation in physical devices have been extensively examined, now is the right time to revisit this understanding. In this paper, we introduce an abstract qubit model (AQM), offering a mathematical framework for higher-level algorithms and applications, and setting forth criteria for lower-level physical devices to enable quantum computation. We first provide a comprehensive definition of "qubits", regarded as the foundational principle for quantum computing algorithms (bottom-up support), and examine their requisites for devices (top-down demand). We then investigate the feasibility of relaxing specific requirements, thereby broadening device support while considering techniques that tradeoff extra costs to counterbalance this relaxation. Lastly, we delve into the quantum applications that only require partial support of "qubits", and discuss the physical systems with limited support of the AQM but remain valuable in quantum applications. AQM may serve as an intermediate interface between quantum algorithms and devices, facilitating quantum algorithm-device co-design.

quant-ph↗

arXiv2024

MP-DPD: Low-Complexity Mixed-Precision Neural Networks for Energy-Efficient Digital Predistortion of Wideband Power Amplifiers

Digital Pre-Distortion (DPD) enhances signal quality in wideband RF power amplifiers (PAs). As signal bandwidths expand in modern radio systems, DPD's energy consumption increasingly impacts overall system efficiency. Deep Neural Networks (DNNs) offer promising advancements in DPD, yet their high complexity hinders their practical deployment. This paper introduces open-source mixed-precision (MP) neural networks that employ quantized low-precision fixed-point parameters for energy-efficient DPD. This approach reduces computational complexity and memory footprint, thereby lowering power consumption without compromising linearization efficacy. Applied to a 160MHz-BW 1024-QAM OFDM signal from a digital RF PA, MP-DPD gives no performance loss against 32-bit floating-point precision DPDs, while achieving -43.75 (L)/-45.27 (R) dBc in Adjacent Channel Power Ratio (ACPR) and -38.72 dB in Error Vector Magnitude (EVM). A 16-bit fixed-point-precision MP-DPD enables a 2.8X reduction in estimated inference power. The PyTorch learning and testing code is publicly available at \url{https://github.com/lab-emi/OpenDPD}.

eess.SP↗

arXiv2024

Stereo-LiDAR Depth Estimation with Deformable Propagation and Learned Disparity-Depth Conversion

Accurate and dense depth estimation with stereo cameras and LiDAR is an important task for automatic driving and robotic perception. While sparse hints from LiDAR points have improved cost aggregation in stereo matching, their effectiveness is limited by the low density and non-uniform distribution. To address this issue, we propose a novel stereo-LiDAR depth estimation network with Semi-Dense hint Guidance, named SDG-Depth. Our network includes a deformable propagation module for generating a semi-dense hint map and a confidence map by propagating sparse hints using a learned deformable window. These maps then guide cost aggregation in stereo matching. To reduce the triangulation error in depth recovery from disparity, especially in distant regions, we introduce a disparity-depth conversion module. Our method is both accurate and efficient. The experimental results on benchmark tests show its superior performance. Our code is available at https://github.com/SJTU-ViSYS/SDG-Depth.

cs.CV↗

arXiv2024

MAPSeg: Unified Unsupervised Domain Adaptation for Heterogeneous Medical Image Segmentation Based on 3D Masked Autoencoding and Pseudo-Labeling

Robust segmentation is critical for deriving quantitative measures from large-scale, multi-center, and longitudinal medical scans. Manually annotating medical scans, however, is expensive and labor-intensive and may not always be available in every domain. Unsupervised domain adaptation (UDA) is a well-studied technique that alleviates this label-scarcity problem by leveraging available labels from another domain. In this study, we introduce Masked Autoencoding and Pseudo-Labeling Segmentation (MAPSeg), a $\textbf{unified}$ UDA framework with great versatility and superior performance for heterogeneous and volumetric medical image segmentation. To the best of our knowledge, this is the first study that systematically reviews and develops a framework to tackle four different domain shifts in medical image segmentation. More importantly, MAPSeg is the first framework that can be applied to $\textbf{centralized}$, $\textbf{federated}$, and $\textbf{test-time}$ UDA while maintaining comparable performance. We compare MAPSeg with previous state-of-the-art methods on a private infant brain MRI dataset and a public cardiac CT-MRI dataset, and MAPSeg outperforms others by a large margin (10.5 Dice improvement on the private MRI dataset and 5.7 on the public CT-MRI dataset). MAPSeg poses great practical value and can be applied to real-world problems. GitHub: https://github.com/XuzheZ/MAPSeg/.

cs.CV↗

arXiv2024

From Graph to Word Bag: Introducing Domain Knowledge to Confusing Charge Prediction

Confusing charge prediction is a challenging task in legal AI, which involves predicting confusing charges based on fact descriptions. While existing charge prediction methods have shown impressive performance, they face significant challenges when dealing with confusing charges, such as Snatch and Robbery. In the legal domain, constituent elements play a pivotal role in distinguishing confusing charges. Constituent elements are fundamental behaviors underlying criminal punishment and have subtle distinctions among charges. In this paper, we introduce a novel From Graph to Word Bag (FWGB) approach, which introduces domain knowledge regarding constituent elements to guide the model in making judgments on confusing charges, much like a judge's reasoning process. Specifically, we first construct a legal knowledge graph containing constituent elements to help select keywords for each charge, forming a word bag. Subsequently, to guide the model's attention towards the differentiating information for each charge within the context, we expand the attention mechanism and introduce a new loss function with attention supervision through words in the word bag. We construct the confusing charges dataset from real-world judicial documents. Experiments demonstrate the effectiveness of our method, especially in maintaining exceptional performance in imbalanced label distributions.

cs.AI↗

arXiv2024

Thermal transport in a 2D amorphous material

Two-dimensional (2D) crystals proved revolutionary soon after graphene was discovered in 2004. However, 2D amorphous materials only became accessible in 2020 and remain largely unexplored. In particular, the thermophysical properties of amorphous materials are of great interest upon transition from 3D to 2D. Here, we probe thermal transport in 2D amorphous carbon. A cross-plane thermal conductivity ($κ$) down to 0.079 $\rm{Wm}^{-1}K^{-1}$ is measured for van der Waals stacked multilayers at room temperature, which is among the lowest reported to date. Meanwhile, an unexpectedly high in-plane $κ$ is obtained for freestanding monolayers which is a few times larger than what is predicted by conventional wisdom for 3D amorphous carbon with similar $\rm{sp}^{2}$ fraction. Our molecular dynamics simulations reveal the role of disorder and highlight the impact of dimensionality. Amorphous materials at the 2D limit open up new avenues for understanding and manipulating heat at the atomic scale.

cond-mat.mtrl-sci↗