arXiv subjects

Ang Li

Publications and source records attributed to Ang Li.

At least 271 records · Page 15Linked to original sources

arXiv2025

Tactic: Adaptive Sparse Attention with Clustering and Distribution Fitting for Long-Context LLMs

Long-context models are essential for many applications but face inefficiencies in loading large KV caches during decoding. Prior methods enforce fixed token budgets for sparse attention, assuming a set number of tokens can approximate full attention. However, these methods overlook variations in the importance of attention across heads, layers, and contexts. To address these limitations, we propose Tactic, a sparsity-adaptive and calibration-free sparse attention mechanism that dynamically selects tokens based on their cumulative attention scores rather than a fixed token budget. By setting a target fraction of total attention scores, Tactic ensures that token selection naturally adapts to variations in attention sparsity. To efficiently approximate this selection, Tactic leverages clustering-based sorting and distribution fitting, allowing it to accurately estimate token importance with minimal computational overhead. We show that Tactic outperforms existing sparse attention algorithms, achieving superior accuracy and up to 7.29x decode attention speedup. This improvement translates to an overall 1.58x end-to-end inference speedup, making Tactic a practical and effective solution for long-context LLM inference in accuracy-sensitive applications.

cs.LG↗

arXiv2025

PSR J1231-1411 revisited: Pulse Profile Analysis of X-ray Observation

One of the primary goals of Neutron Star Interior Composition Explorer (NICER)-like X-ray missions is to impose stringent constraints on the neutron star equation of state by precisely measuring their masses and radii. NICER has recently expanded the dataset of inferred mass-radius relations for neutron stars, including four rotation-powered millisecond pulsars PSR J0030+0451, PSR J0740+6620, PSR J0437-4715, and PSR J1231-1411. In this work, the mass-radius relation and X-ray emitting region properties of PSR J1231-1411 are inferred with an independent pulse profile modeling based on the spherical star Schwarzschild-spacetime and Doppler approximation. With one single-temperature elongated hot spot and one single-temperature crescent hot spot, the inferred gravitational mass is $M = 1.12 \pm 0.07 M_{\odot}$ and the inferred equatorial radius is $R_{eq} = 9.91_{-0.86}^{+0.88}$ km (68% credible intervals). It provides an alternative geometry configuration of the X-ray emitting region for PSR J1231-1411 to sufficiently explain the observation data of NICER and XMM-Newton. The inferred radius is smaller than that derived by \citet{salmi2024nicer} ($M = 1.04_{-0.03}^{+0.05} M_{\odot}$, $R_{eq} = 12.6 \pm 0.3$ km), and the inferred mass is slightly higher in this work. The inferred geometry configurations of the X-ray emitting region in both works are non-antipodal, which is not consistent with a centered dipole magnetic field and suggests a complex magnetic field structure.

astro-ph.HE↗

arXiv2025

Estimating Probabilities of Causation with Machine Learning Models

Probabilities of causation play a crucial role in modern decision-making. This paper addresses the challenge of predicting probabilities of causation for subpopulations with insufficient data using machine learning models. Tian and Pearl first defined and derived tight bounds for three fundamental probabilities of causation: the probability of necessity and sufficiency (PNS), the probability of sufficiency (PS), and the probability of necessity (PN). However, estimating these probabilities requires both experimental and observational distributions specific to each subpopulation, which are often unavailable or impractical to obtain with limited population-level data. We assume that the probabilities of causation for each subpopulation are determined by its characteristics. To estimate these probabilities for subpopulations with insufficient data, we propose using machine learning models that draw insights from subpopulations with sufficient data. Our evaluation of multiple machine learning models indicates that, given sufficient population-level data and an appropriate choice of machine learning model and activation function, PNS can be effectively predicted. Through simulation studies, we show that our multilayer perceptron (MLP) model with the Mish activation function achieves a mean absolute error (MAE) of approximately 0.02 in predicting PNS for 32,768 subpopulations using data from around 2,000 subpopulations.

cs.AI↗

arXiv2025

Inclusive Avatar Guidelines for People with Disabilities: Supporting Disability Representation in Social Virtual Reality

Avatar is a critical medium for identity representation in social virtual reality (VR). However, options for disability expression are highly limited on current avatar interfaces. Improperly designed disability features may even perpetuate misconceptions about people with disabilities (PWD). As more PWD use social VR, there is an emerging need for comprehensive design standards that guide developers and designers to create inclusive avatars. Our work aim to advance the avatar design practices by delivering a set of centralized, comprehensive, and validated design guidelines that are easy to adopt, disseminate, and update. Through a systematic literature review and interview with 60 participants with various disabilities, we derived 20 initial design guidelines that cover diverse disability expression methods through five aspects, including avatar appearance, body dynamics, assistive technology design, peripherals around avatars, and customization control. We further evaluated the guidelines via a heuristic evaluation study with 10 VR practitioners, validating the guideline coverage, applicability, and actionability. Our evaluation resulted in a final set of 17 design guidelines with recommendation levels.

cs.HC↗

arXiv2025

Commercial LLM Agents Are Already Vulnerable to Simple Yet Dangerous Attacks

A high volume of recent ML security literature focuses on attacks against aligned large language models (LLMs). These attacks may extract private information or coerce the model into producing harmful outputs. In real-world deployments, LLMs are often part of a larger agentic pipeline including memory systems, retrieval, web access, and API calling. Such additional components introduce vulnerabilities that make these LLM-powered agents much easier to attack than isolated LLMs, yet relatively little work focuses on the security of LLM agents. In this paper, we analyze security and privacy vulnerabilities that are unique to LLM agents. We first provide a taxonomy of attacks categorized by threat actors, objectives, entry points, attacker observability, attack strategies, and inherent vulnerabilities of agent pipelines. We then conduct a series of illustrative attacks on popular open-source and commercial agents, demonstrating the immediate practical implications of their vulnerabilities. Notably, our attacks are trivial to implement and require no understanding of machine learning.

cs.LG↗

arXiv2025

DPD-NeuralEngine: A 22-nm 6.6-TOPS/W/mm$^2$ Recurrent Neural Network Accelerator for Wideband Power Amplifier Digital Pre-Distortion

The increasing adoption of Deep Neural Network (DNN)-based Digital Pre-distortion (DPD) in modern communication systems necessitates efficient hardware implementations. This paper presents DPD-NeuralEngine, an ultra-fast, tiny-area, and power-efficient DPD accelerator based on a Gated Recurrent Unit (GRU) neural network (NN). Leveraging a co-designed software and hardware approach, our 22 nm CMOS implementation operates at 2 GHz, capable of processing I/Q signals up to 250 MSps. Experimental results demonstrate a throughput of 256.5 GOPS and power efficiency of 1.32 TOPS/W with DPD linearization performance measured in Adjacent Channel Power Ratio (ACPR) of -45.3 dBc and Error Vector Magnitude (EVM) of -39.8 dB. To our knowledge, this work represents the first AI-based DPD application-specific integrated circuit (ASIC) accelerator, achieving a power-area efficiency (PAE) of 6.6 TOPS/W/mm$^2$.

cs.AR↗

arXiv2025

Mitigating Biases of Large Language Models in Stance Detection with Counterfactual Augmented Calibration

Stance detection is critical for understanding the underlying position or attitude expressed toward a topic. Large language models (LLMs) have demonstrated significant advancements across various natural language processing tasks including stance detection, however, their performance in stance detection is limited by biases and spurious correlations inherent due to their data-driven nature. Our statistical experiment reveals that LLMs are prone to generate biased stances due to sentiment-stance spurious correlations and preference towards certain individuals and topics. Furthermore, the results demonstrate a strong negative correlation between stance bias and stance detection performance, underscoring the importance of mitigating bias to enhance the utility of LLMs in stance detection. Therefore, in this paper, we propose a Counterfactual Augmented Calibration Network (FACTUAL), which a novel calibration network is devised to calibrate potential bias in the stance prediction of LLMs. Further, to address the challenge of effectively learning bias representations and the difficulty in the generalizability of debiasing, we construct counterfactual augmented data. This approach enhances the calibration network, facilitating the debiasing and out-of-domain generalization. Experimental results on in-target and zero-shot stance detection tasks show that the proposed FACTUAL can effectively mitigate biases of LLMs, achieving state-of-the-art results.

cs.CL↗

arXiv2025

Electric-field Switching of Interlayer Magnetic Order in a van der Waals Heterobilayer via Spin-potential Coupling

Electric-field switching of magnetic order is of significant physical interest and holds great potential for spintronic applications. However, it has rarely been reported in two-dimensional (2D) van der Waals (vdW) magnets due to the inherently weak interaction between spin order and electric fields. Here we propose a general spin-potential mechanism that significantly enhances the magnetoelectric coupling. As a result, the relative stability of different interlayer magnetic orders in an asymmetric van der Waals heterobilayer can be reversed by external electric fields via spin-potential coupling. Based on this mechanism, we designed a series of 2D vdW all-magnetic heterobilayers, such as CrI3/MnSe2, in which a transition from interlayer ferromagnetic (iFM) to antiferromagnetic (iAFM) order is realized by a feasible electric field around 0.1 V/Å. Our findings not only reveal a novel magnetoelectric coupling mechanism, but also present a practical strategy for achieving pure electric-field-driven magnetic order switching

cond-mat.mtrl-sci↗

arXiv2025

A Tensor-Train Decomposition based Compression of LLMs on Group Vector Systolic Accelerator

Large language models (LLMs) are both storage-intensive and computation-intensive, posing significant challenges when deployed on resource-constrained hardware. As linear layers in LLMs are mainly resource consuming parts, this paper develops a tensor-train decomposition (TTD) for LLMs with a further hardware implementation on FPGA. TTD compression is applied to the linear layers in ChatGLM3-6B and LLaMA2-7B models with compression ratios (CRs) for the whole network 1.94$\times$ and 1.60$\times$, respectively. The compressed LLMs are further implemented on FPGA hardware within a highly efficient group vector systolic array (GVSA) architecture, which has DSP-shared parallel vector PEs for TTD inference, as well as optimized data communication in the accelerator. Experimental results show that the corresponding TTD based LLM accelerator implemented on FPGA achieves 1.45$\times$ and 1.57$\times$ reduction in first token delay for ChatGLM3-6B and LLaMA2-7B models, respectively.

cs.AR↗

arXiv2025

ATTNChecker: Highly-Optimized Fault Tolerant Attention for Large Language Model Training

Large Language Models (LLMs) have demonstrated remarkable performance in various natural language processing tasks. However, the training of these models is computationally intensive and susceptible to faults, particularly in the attention mechanism, which is a critical component of transformer-based LLMs. In this paper, we investigate the impact of faults on LLM training, focusing on INF, NaN, and near-INF values in the computation results with systematic fault injection experiments. We observe the propagation patterns of these errors, which can trigger non-trainable states in the model and disrupt training, forcing the procedure to load from checkpoints. To mitigate the impact of these faults, we propose ATTNChecker, the first Algorithm-Based Fault Tolerance (ABFT) technique tailored for the attention mechanism in LLMs. ATTNChecker is designed based on fault propagation patterns of LLM and incorporates performance optimization to adapt to both system reliability and model vulnerability while providing lightweight protection for fast LLM training. Evaluations on four LLMs show that ATTNChecker incurs on average 7% overhead on training while detecting and correcting all extreme errors. Compared with the state-of-the-art checkpoint/restore approach, ATTNChecker reduces recovery overhead by up to 49x.

cs.DC↗

arXiv2025

Science objectives of the Einstein Probe mission

The Einstein Probe (EP) is an interdisciplinary mission of time-domain and X-ray astronomy. Equipped with a wide-field lobster-eye X-ray focusing imager, EP will discover cosmic X-ray transients and monitor the X-ray variability of known sources in 0.5-4 keV, at a combination of detecting sensitivity and cadence that is not accessible to the previous and current wide-field monitoring missions. EP can perform quick characterisation of transients or outbursts with a Wolter-I X-ray telescope onboard. In this paper, the science objectives of the Einstein Probe mission are presented. EP is expected to enlarge the sample of previously known or predicted but rare types of transients with a wide range of timescales. Among them, fast extragalactic transients will be surveyed systematically in soft X-rays, which include γ-ray bursts and their variants, supernova shock breakouts, and the predicted X-ray transients associated with binary neutron star mergers. EP will detect X-ray tidal disruption events and outbursts from active galactic nuclei, possibly at an early phase of the flares for some. EP will monitor the variability and outbursts of X-rays from white dwarfs, neutron stars and black holes in our and neighbouring galaxies at flux levels fainter than those detectable by the current instruments, and is expected to discover new objects. A large sample of stellar X-ray flares will also be detected and characterised. In the era of multi-messenger astronomy, EP has the potential of detecting the possible X-ray counterparts of gravitational wave events, neutrino sources, and ultra-high energy γ-ray and cosmic ray sources. EP is expected to help advance the studies of extreme objects/phenomena and their underlying physical processes revealed in the dynamic X-ray universe, as well as studies in other areas of X-ray astronomy.

astro-ph.HE↗

arXiv2024

Early Exploration of a Flexible Framework for Efficient Quantum Linear Solvers in Power Systems

The rapid integration of renewable energy resources presents formidable challenges in managing power grids. While advanced computing and machine learning techniques offer some solutions for accelerating grid modeling and simulation, there remain complex problems that classical computers cannot effectively address. Quantum computing, a promising technology, has the potential to fundamentally transform how we manage power systems, especially in scenarios with a higher proportion of renewable energy sources. One critical aspect is solving large-scale linear systems of equations, crucial for power system applications like power flow analysis, for which the Harrow-Hassidim-Lloyd (HHL) algorithm is a well-known quantum solution. However, HHL quantum circuits often exhibit excessive depth, making them impractical for current Noisy-Intermediate-Scale-Quantum (NISQ) devices. In this paper, we introduce a versatile framework, powered by NWQSim, that bridges the gap between power system applications and quantum linear solvers available in Qiskit. This framework empowers researchers to efficiently explore power system applications using quantum linear solvers. Through innovative gate fusion strategies, reduced circuit depth, and GPU acceleration, our simulator significantly enhances resource efficiency. Power flow case studies have demonstrated up to a eight-fold speedup compared to Qiskit Aer, all while maintaining comparable levels of accuracy.

quant-ph↗

arXiv2024

Model-GLUE: Democratized LLM Scaling for A Large Model Zoo in the Wild

As Large Language Models (LLMs) excel across tasks and specialized domains, scaling LLMs based on existing models has garnered significant attention, which faces the challenge of decreasing performance when combining disparate models. Various techniques have been proposed for the aggregation of pre-trained LLMs, including model merging, Mixture-of-Experts, and stacking. Despite their merits, a comprehensive comparison and synergistic application of them to a diverse model zoo is yet to be adequately addressed. In light of this research gap, this paper introduces Model-GLUE, a holistic LLM scaling guideline. First, our work starts with a benchmarking of existing LLM scaling techniques, especially selective merging, and variants of mixture. Utilizing the insights from the benchmark results, we formulate an optimal strategy for the selection and aggregation of a heterogeneous model zoo characterizing different architectures and initialization.Our methodology involves the clustering of mergeable models and optimal merging strategy selection, and the integration of clusters through a model mixture. Finally, evidenced by our experiments on a diverse Llama-2-based model zoo, Model-GLUE shows an average performance enhancement of 5.61%, achieved without additional training. Codes are available at: https://github.com/Model-GLUE/Model-GLUE.

cs.LG↗

arXiv2024

SymBreak: Mitigating Quantum Degeneracy Issues in QLDPC Code Decoders by Breaking Symmetry

Quantum error correction (QEC) is critical for scalable and reliable quantum computing, but existing solutions, such as surface codes, incur significant qubit overhead. Quantum low-density parity check (qLDPC) codes have recently emerged as a promising alternative, requiring fewer qubits. However, the lack of efficient decoders remains a major barrier to their practical implementation. In this work, we introduce SymBreak, a novel decoder for qLDPC codes that adaptively modifies the decoding graph to improve the performance of state-of-the-art belief propagation (BP) decoders. Our key contribution is identifying quantum degeneracy as a root cause of the convergence issues often encountered in BP decoding of quantum LDPC codes. We propose a solution that mitigates this issue at the decoding graph level, achieving both fast and accurate decoding. Our results demonstrate that SymBreak outperforms BP and BP+OSD-a more complex variant of BP-with a $16.17\times$ reduction in logical error rate compared to BP and $3.23\times$ compared to BP+OSD across various qLDPC code families. With only an $18.97$% time overhead compared to BP, SymBreak provides significantly faster decoding times than BP+OSD, representing a major advancement in efficient and accurate decoding for qLDPC-based QEC architectures.

quant-ph↗

arXiv2024

CaliScalpel: In-Situ and Fine-Grained Qubit Calibration Integrated with Surface Code Quantum Error Correction

Quantum Error Correction (QEC) is a cornerstone of fault-tolerant, large-scale quantum computing. However, qubit error drift significantly degrades QEC performance over time, necessitating periodic calibration. Traditional calibration methods disrupt quantum states, requiring system downtime and making in situ calibration infeasible. We present CaliScalpel, an innovative framework for in situ calibration in surface codes. The core idea behind CaliScalpel is leveraging code deformation to isolate qubits undergoing calibration from logical patches. This allows calibration to proceed concurrently with computation, while code enlargement maintains error correction capabilities with minimal qubit overhead. Additionally, CaliScalpel incorporates optimized calibration schedules derived from detailed device characterization, effectively minimizing physical error rates. Our results show that CaliScalpel achieves concurrent calibration and computation with modest qubit overhead and negligible execution time impact, marking a significant step toward practical in situ calibration in surface-code-based quantum computing systems.

quant-ph↗

arXiv2024

MERIT: Multimodal Wearable Vital Sign Waveform Monitoring

Cardiovascular disease (CVD) is the leading cause of death and premature mortality worldwide, with occupational environments significantly influencing CVD risk, underscoring the need for effective cardiac monitoring and early warning systems. Existing methods of monitoring vital signs require subjects to remain stationary, which is impractical for daily monitoring as individuals are often in motion. To address this limitation, we propose MERIT, a multimodality-based wearable system designed for precise ECG waveform monitoring without movement restrictions. Daily activities, involving frequent arm movements, can significantly affect sensor data and complicate the reconstruction of accurate ECG signals. To mitigate motion impact and enhance ECG signal reconstruction, we introduce a deep independent component analysis (Deep-ICA) module and a multimodal fusion module. We conducted experiments with 15 subjects. Our results, compared with commercial wearable devices and existing methods, demonstrate that MERIT accurately reconstructs ECG waveforms during various office activities, offering a reliable solution for fine-grained cardiac monitoring in dynamic environments.

eess.SY↗

arXiv2024

t-READi: Transformer-Powered Robust and Efficient Multimodal Inference for Autonomous Driving

Given the wide adoption of multimodal sensors (e.g., camera, lidar, radar) by autonomous vehicles (AVs), deep analytics to fuse their outputs for a robust perception become imperative. However, existing fusion methods often make two assumptions rarely holding in practice: i) similar data distributions for all inputs and ii) constant availability for all sensors. Because, for example, lidars have various resolutions and failures of radars may occur, such variability often results in significant performance degradation in fusion. To this end, we present tREADi, an adaptive inference system that accommodates the variability of multimodal sensory data and thus enables robust and efficient perception. t-READi identifies variation-sensitive yet structure-specific model parameters; it then adapts only these parameters while keeping the rest intact. t-READi also leverages a cross-modality contrastive learning method to compensate for the loss from missing modalities. Both functions are implemented to maintain compatibility with existing multimodal deep fusion methods. The extensive experiments evidently demonstrate that compared with the status quo approaches, t-READi not only improves the average inference accuracy by more than 6% but also reduces the inference latency by almost 15x with the cost of only 5% extra memory overhead in the worst case under realistic data and modal variations.

cs.CV↗

arXiv2024

PowerMove: Optimizing Compilation for Neutral Atom Quantum Computers with Zoned Architecture

Neutral atom-based quantum computers (NAQCs) have recently emerged as promising candidates for scalable quantum computing, largely due to their advanced hardware capabilities, particularly qubit movement and the zoned architecture (ZA). However, fully leveraging these features poses significant compiler challenges, as it requires addressing complexities across gate scheduling, qubit allocation, qubit movement, and inter-zone communication. In this paper, we present PowerMove, an efficient compiler for NAQCs that enhances the qubit movement framework while fully integrating the advantages of ZA. By recognizing and leveraging the interdependencies between these key aspects, PowerMove unlocks new optimization opportunities, significantly enhancing both scalability and fidelity. Our evaluation demonstrates an improvement in fidelity by several orders of magnitude compared to the state-of-the-art methods, with execution time improved by up to 3.46x and compilation time reduced by up to 213.5x. We will open-source our code later to foster further research and collaboration within the community.

quant-ph↗