arXiv subjects

Jue Wang

Publications and source records attributed to Jue Wang.

At least 73 records · Page 4Linked to original sources

arXiv2025

CHASe: Client Heterogeneity-Aware Data Selection for Effective Federated Active Learning

Active learning (AL) reduces human annotation costs for machine learning systems by strategically selecting the most informative unlabeled data for annotation, but performing it individually may still be insufficient due to restricted data diversity and annotation budget. Federated Active Learning (FAL) addresses this by facilitating collaborative data selection and model training, while preserving the confidentiality of raw data samples. Yet, existing FAL methods fail to account for the heterogeneity of data distribution across clients and the associated fluctuations in global and local model parameters, adversely affecting model accuracy. To overcome these challenges, we propose CHASe (Client Heterogeneity-Aware Data Selection), specifically designed for FAL. CHASe focuses on identifying those unlabeled samples with high epistemic variations (EVs), which notably oscillate around the decision boundaries during training. To achieve both effectiveness and efficiency, \model{} encompasses techniques for 1) tracking EVs by analyzing inference inconsistencies across training epochs, 2) calibrating decision boundaries of inaccurate models with a new alignment loss, and 3) enhancing data selection efficiency via a data freeze and awaken mechanism with subset sampling. Experiments show that CHASe surpasses various established baselines in terms of effectiveness and efficiency, validated across diverse datasets, model complexities, and heterogeneous federation settings.

cs.LG↗

arXiv2025

HMI: Hierarchical Knowledge Management for Efficient Multi-Tenant Inference in Pretrained Language Models

The significant computational demands of pretrained language models (PLMs), which often require dedicated hardware, present a substantial challenge in serving them efficiently, especially in multi-tenant environments. To address this, we introduce HMI, a Hierarchical knowledge management-based Multi-tenant Inference system, designed to manage tenants with distinct PLMs resource-efficiently. Our approach is three-fold: Firstly, we categorize PLM knowledge into general, domain-specific, and task-specific. Leveraging insights on knowledge acquisition across different model layers, we construct hierarchical PLMs (hPLMs) by extracting and storing knowledge at different levels, significantly reducing GPU memory usage per tenant. Secondly, we establish hierarchical knowledge management for hPLMs generated by various tenants in HMI. We manage domain-specific knowledge with acceptable storage increases by constructing and updating domain-specific knowledge trees based on frequency. We manage task-specific knowledge within limited GPU memory through parameter swapping. Finally, we propose system optimizations to enhance resource utilization and inference throughput. These include fine-grained pipelining via hierarchical knowledge prefetching to overlap CPU and I/O operations with GPU computations, and optimizing parallel implementations with batched matrix multiplications. Our experimental results demonstrate that the proposed HMI can efficiently serve up to 10,000 hPLMs (hBERTs and hGPTs) on a single GPU, with only a negligible compromise in accuracy.

cs.LG↗

arXiv2025

Why We Feel: Breaking Boundaries in Emotional Reasoning with Multimodal Large Language Models

Most existing emotion analysis emphasizes which emotion arises (e.g., happy, sad, angry) but neglects the deeper why. We propose Emotion Interpretation (EI), focusing on causal factors-whether explicit (e.g., observable objects, interpersonal interactions) or implicit (e.g., cultural context, off-screen events)-that drive emotional responses. Unlike traditional emotion recognition, EI tasks require reasoning about triggers instead of mere labeling. To facilitate EI research, we present EIBench, a large-scale benchmark encompassing 1,615 basic EI samples and 50 complex EI samples featuring multifaceted emotions. Each instance demands rationale-based explanations rather than straightforward categorization. We further propose a Coarse-to-Fine Self-Ask (CFSA) annotation pipeline, which guides Vision-Language Models (VLLMs) through iterative question-answer rounds to yield high-quality labels at scale. Extensive evaluations on open-source and proprietary large language models under four experimental settings reveal consistent performance gaps-especially for more intricate scenarios-underscoring EI's potential to enrich empathetic, context-aware AI applications. Our benchmark and methods are publicly available at: https://github.com/Lum1104/EIBench, offering a foundation for advanced multimodal causal analysis and next-generation affective computing.

cs.AI↗

arXiv2025

Scaling Instruction-Tuned LLMs to Million-Token Contexts via Hierarchical Synthetic Data Generation

Large Language Models (LLMs) struggle with long-context reasoning, not only due to the quadratic scaling of computational complexity with sequence length but also because of the scarcity and expense of annotating long-context data. There has been barely any open-source work that systematically ablates long-context data, nor is there any openly available instruction tuning dataset with contexts surpassing 100K tokens. To bridge this gap, we introduce a novel post-training synthetic data generation strategy designed to efficiently extend the context window of LLMs while preserving their general task performance. Our approach scalably extends to arbitrarily long context lengths, unconstrained by the length of available real-world data, which effectively addresses the scarcity of raw long-context data. Through a step-by-step rotary position embedding (RoPE) scaling training strategy, we demonstrate that our model, with a context length of up to 1M tokens, performs well on the RULER benchmark and InfiniteBench and maintains robust performance on general language tasks.

cs.CL↗

arXiv2025

Neutron particle transport 3D method of characteristic Multi GPU platform Parallel Computing

Three-dimensional neutron transport calculations using the Method of Characteristics (MOC) are highly regarded for their exceptional computational efficiency, precision, and stability. Nevertheless, when dealing with extensive-scale computations, the computational demands are substantial, leading to prolonged computation times. To address this challenge while considering GPU memory limitations, this study transplants the real-time generation and characteristic line computation techniques onto the GPU platform. Empirical evidence emphasizes that the GPU-optimized approach maintains a heightened level of precision in computation results and produces a significant acceleration effect. Furthermore, to fully harness the computational capabilities of GPUs, a dual approach involving characteristic line preloading and load balancing mechanisms is adopted, further enhancing computational efficiency. The resulting increase in computational efficiency, compared to traditional methods, reaches an impressive 300 to 400-fold improvement.

cs.DC↗

arXiv2025

Train Small, Infer Large: Memory-Efficient LoRA Training for Large Language Models

Large Language Models (LLMs) have significantly advanced natural language processing with exceptional task generalization capabilities. Low-Rank Adaption (LoRA) offers a cost-effective fine-tuning solution, freezing the original model parameters and training only lightweight, low-rank adapter matrices. However, the memory footprint of LoRA is largely dominated by the original model parameters. To mitigate this, we propose LoRAM, a memory-efficient LoRA training scheme founded on the intuition that many neurons in over-parameterized LLMs have low training utility but are essential for inference. LoRAM presents a unique twist: it trains on a pruned (small) model to obtain pruned low-rank matrices, which are then recovered and utilized with the original (large) model for inference. Additionally, minimal-cost continual pre-training, performed by the model publishers in advance, aligns the knowledge discrepancy between pruned and original models. Our extensive experiments demonstrate the efficacy of LoRAM across various pruning strategies and downstream tasks. For a model with 70 billion parameters, LoRAM enables training on a GPU with only 20G HBM, replacing an A100-80G GPU for LoRA training and 15 GPUs for full fine-tuning. Specifically, QLoRAM implemented by structured pruning combined with 4-bit quantization, for LLaMA-3.1-70B (LLaMA-2-70B), reduces the parameter storage cost that dominates the memory usage in low-rank matrix training by 15.81$\times$ (16.95$\times$), while achieving dominant performance gains over both the original LLaMA-3.1-70B (LLaMA-2-70B) and LoRA-trained LLaMA-3.1-8B (LLaMA-2-13B). Code is available at https://github.com/junzhang-zj/LoRAM.

cs.LG↗

arXiv2025

HOPS: High-order Polynomials with Self-supervised Dimension Reduction for Load Forecasting

Load forecasting is a fundamental task in smart grid. Many techniques have been applied to developing load forecasting models. Due to the challenges such as the Curse of Dimensionality, overfitting, and limited computing resources, multivariate higher-order polynomial models have received limited attention in load forecasting, despite their desirable mathematical foundations and optimization properties. In this paper, we propose low rank approximation and self-supervised dimension reduction to address the aforementioned issues. To further improve computational efficiency, we also utilize a fast Conjugate Gradient based algorithm for the proposed polynomial models. Based on the load datasets from the ISO New England, the proposed method high-order polynomials with self-supervised dimension reduction (HOPS) demonstrates higher forecasting accuracy over several competitive models. Additionally, experimental results indicate that our approach alleviates redundant variable construction, achieving better forecasts with fewer input variables.

cs.LG↗

arXiv2025

Diffusion Approximation for Slow-Fast SDEs with State-Dependent Switching

In this paper, we study the diffusion approximation for slow-fast stochastic differential equations with state-dependent switching, where the slow component $X^{\varepsilon}$ is the solution of a stochastic differential equation with additional homogenization term, while the fast component $α^{\varepsilon}$ is a switching process. We first prove the weak convergence of $\{X^\varepsilon\}_{0<\varepsilon\leq 1}$ to $\bar{X}$ in the space of continuous functions, as $\varepsilon\rightarrow 0$. Using the martingale problem approach and Poisson equation associated with a Markov chain, we identify this weak limiting process as the unique solution $\bar{X}$ of a new stochastic differential equation, which has new drift and diffusion terms that differ from those in the original equation. Next, we prove the order $1/2$ of weak convergence of $X^{\varepsilon}_t$ to $\bar{X}_t$ by applying suitable test functions $ϕ$, for any $t\in [0, T]$. Additionally, we provide an example to illustrate that the order we achieve is optimal.

math.PR↗

arXiv2025

Channel Gain Map Construction based on Subregional Learning and Prediction

The construction of channel gain map (CGM) is essential for realizing environment-aware wireless communications expected in 6G, for which a fundamental problem is how to predict the channel gains at unknown locations effectively by a finite number of measurements. As using a single prediction model is not effective in complex propagation environments, we propose a subregional learning-based CGM construction scheme, with which the entire map is divided into subregions via data-driven clustering, then individual models are constructed and trained for every subregion. In this way, specific propagation feature in each subregion can be better extracted with finite training data. Moreover, we propose to further improve prediction accuracy by uneven subregion sampling, as well as training data reuse around the subregion boundaries. Simulation results validate the effectiveness of the proposed scheme in CGM construction.

cs.NI↗

arXiv2025

Acc-SpMM: Accelerating General-purpose Sparse Matrix-Matrix Multiplication with GPU Tensor Cores

General-purpose Sparse Matrix-Matrix Multiplication (SpMM) is a fundamental kernel in scientific computing and deep learning. The emergence of new matrix computation units such as Tensor Cores (TCs) brings more opportunities for SpMM acceleration. However, in order to fully unleash the power of hardware performance, systematic optimization is required. In this paper, we propose Acc-SpMM, a high-performance SpMM library on TCs, with multiple optimizations, including data-affinity-based reordering, memory efficient compressed format, high-throughput pipeline, and adaptive sparsity-aware load balancing. In contrast to the state-of-the-art SpMM kernels on various NVIDIA GPU architectures with a diverse range of benchmark matrices, Acc-SpMM achieves significant performance improvements, on average 2.52x (up to 5.11x) speedup on RTX 4090, on average 1.91x (up to 4.68x) speedup on A800, and on average 1.58x (up to 3.60x) speedup on H100 over cuSPARSE.

cs.DC↗

arXiv2025

Heterogeneous-free narrow linewidth semiconductor laser with optical injection locking

Narrow linewidth lasers are indispensable for coherent optical systems, including communications, metrology, and sensing. Although compact semiconductor lasers with narrow linewidths and low noise have been demonstrated, their spectral purity typically relies on hybrid or heterogeneous external cavity feedback. Here, we present a theoretical and experimental demonstration of a heterogeneous free optical injection locking (HF OIL) semiconductor laser. By integrating a topological interface state extended (TISE) laser with a micro ring resonator (MRR) on an AlGaInAs multiple quantum well platform,we achieve monolithic photon injection and phase locking, thereby reducing the optical linewidth. We fabricated and characterized a 1550 nm sidewall HF OIL laser, achieving stable single mode operation over a broad current range (65 to 300 mA) and a side mode suppression ratio (SMSR) over 50 dB. Under injection locking, the devices Voigt fitted linewidth narrowed from over 1.7 MHz (free running) to 4.2 kHz, representing a three order of magnitude improvement over conventional distributed feedback lasers. The intrinsic linewidth of 1.4 kHz is measured by correlated delayed self-heterodyne frequency noise power spectrum density (FN PSD) method. Moreover, the HF OIL laser demonstrated high phase stability and the ability to transition from a random phased to a phase locked state. These results underscore the potential of HF-OIL lasers in advancing coherent optical communications and phase encoders in quantum key distribution (QKD) systems.

physics.optics↗

arXiv2024

Channel Charting-assisted Non-orthogonal Pilot Allocation for Uplink XL-MIMO Transmission

Extremely large-scale multiple-input multiple-output (XL-MIMO) is critical to future wireless networks. The substantial increase in the number of base station (BS) antennas introduces near-field propagation effects in the wireless channels, complicating channel parameter estimation and increasing pilot overhead. Channel charting (CC) has emerged as a potent unsupervised technique to effectively harness varying high-dimensional channel statistics to enable non-orthogonal pilot assignment and reduce pilot overhead. In this paper, we investigate near-field channel estimation with reduced pilot overhead by developing a CC-assisted pilot scheduling. To this end, we introduce a polar-domain codebook to capture the power distribution of near-field XL-MIMO channels. The CC-assisted approach uses such features as inputs to enable an effective low-dimensional mapping of the inherent correlation patterns in near-field user terminal (UT) channels. Building upon the mapped channel correlations, we further propose a near-field CC-assisted pilot allocation (NCC-PA) algorithm, which efficiently enhances channel orthogonality among pilot-reusing UTs. Numerical results confirm that the NCC-PA algorithm substantially elevates the wireless transmission performance, offering a marked improvement over the conventional far-field CC-PA approach.

cs.IT↗

arXiv2024

NowYouSee Me: Context-Aware Automatic Audio Description

Audio Description (AD) plays a pivotal role as an application system aimed at guaranteeing accessibility in multimedia content, which provides additional narrations at suitable intervals to describe visual elements, catering specifically to the needs of visually impaired audiences. In this paper, we introduce $\mathrm{CA^3D}$, the pioneering unified Context-Aware Automatic Audio Description system that provides AD event scripts with precise locations in the long cinematic content. Specifically, $\mathrm{CA^3D}$ system consists of: 1) a Temporal Feature Enhancement Module to efficiently capture longer term dependencies, 2) an anchor-based AD event detector with feature suppression module that localizes the AD events and extracts discriminative feature for AD generation, and 3) a self-refinement module that leverages the generated output to tweak AD event boundaries from coarse to fine. Unlike conventional methods which rely on metadata and ground truth AD timestamp for AD detection and generation tasks, the proposed $\mathrm{CA^3D}$ is the first end-to-end trainable system that only uses visual cue. Extensive experiments demonstrate that the proposed $\mathrm{CA^3D}$ improves existing architectures for both AD event detection and script generation metrics, establishing the new state-of-the-art performances in the AD automation.

cs.CV↗

arXiv2024

GEXIA: Granularity Expansion and Iterative Approximation for Scalable Multi-grained Video-language Learning

In various video-language learning tasks, the challenge of achieving cross-modality alignment with multi-grained data persists. We propose a method to tackle this challenge from two crucial perspectives: data and modeling. Given the absence of a multi-grained video-text pretraining dataset, we introduce a Granularity EXpansion (GEX) method with Integration and Compression operations to expand the granularity of a single-grained dataset. To better model multi-grained data, we introduce an Iterative Approximation Module (IAM), which embeds multi-grained videos and texts into a unified, low-dimensional semantic space while preserving essential information for cross-modal alignment. Furthermore, GEXIA is highly scalable with no restrictions on the number of video-text granularities for alignment. We evaluate our work on three categories of video tasks across seven benchmark datasets, showcasing state-of-the-art or comparable performance. Remarkably, our model excels in tasks involving long-form video understanding, even though the pretraining dataset only contains short video clips.

cs.CV↗

arXiv2024

Video Token Merging for Long-form Video Understanding

As the scale of data and models for video understanding rapidly expand, handling long-form video input in transformer-based models presents a practical challenge. Rather than resorting to input sampling or token dropping, which may result in information loss, token merging shows promising results when used in collaboration with transformers. However, the application of token merging for long-form video processing is not trivial. We begin with the premise that token merging should not rely solely on the similarity of video tokens; the saliency of tokens should also be considered. To address this, we explore various video token merging strategies for long-form video classification, starting with a simple extension of image token merging, moving to region-concentrated merging, and finally proposing a learnable video token merging (VTM) algorithm that dynamically merges tokens based on their saliency. Extensive experimental results show that we achieve better or comparable performances on the LVU, COIN, and Breakfast datasets. Moreover, our approach significantly reduces memory costs by 84% and boosts throughput by approximately 6.89 times compared to baseline algorithms.

cs.CV↗

arXiv2024

High-Fidelity GAN Inversion for Image Attribute Editing

We present a novel high-fidelity generative adversarial network (GAN) inversion framework that enables attribute editing with image-specific details well-preserved (e.g., background, appearance, and illumination). We first analyze the challenges of high-fidelity GAN inversion from the perspective of lossy data compression. With a low bit-rate latent code, previous works have difficulties in preserving high-fidelity details in reconstructed and edited images. Increasing the size of a latent code can improve the accuracy of GAN inversion but at the cost of inferior editability. To improve image fidelity without compromising editability, we propose a distortion consultation approach that employs a distortion map as a reference for high-fidelity reconstruction. In the distortion consultation inversion (DCI), the distortion map is first projected to a high-rate latent map, which then complements the basic low-rate latent code with more details via consultation fusion. To achieve high-fidelity editing, we propose an adaptive distortion alignment (ADA) module with a self-supervised training scheme, which bridges the gap between the edited and inversion images. Extensive experiments in the face and car domains show a clear improvement in both inversion and editing quality.

cs.CV↗

arXiv2024

De novo design of high-affinity protein binders with AlphaProteo

Computational design of protein-binding proteins is a fundamental capability with broad utility in biomedical research and biotechnology. Recent methods have made strides against some target proteins, but on-demand creation of high-affinity binders without multiple rounds of experimental testing remains an unsolved challenge. This technical report introduces AlphaProteo, a family of machine learning models for protein design, and details its performance on the de novo binder design problem. With AlphaProteo, we achieve 3- to 300-fold better binding affinities and higher experimental success rates than the best existing methods on seven target proteins. Our results suggest that AlphaProteo can generate binders "ready-to-use" for many research applications using only one round of medium-throughput screening and no further optimization.

q-bio.BM↗

arXiv2024

Optical signatures of interlayer electron coherence in a bilayer semiconductor

Emergent strongly-correlated electronic phenomena in atomically-thin transition metal dichalcogenides are an exciting frontier in condensed matter physics, with examples ranging from bilayer superconductivity~\cite{zhao2023evidence} and electronic Wigner crystals~\cite{smolenski2021signatures,zhou2021bilayer} to the ongoing quest for exciton condensation~\cite{wang2019evidence,ma2021strongly,shi2022bilayer}. Here, we experimentally investigate the properties of indirect excitons in naturally-grown MoS$_2$-homobilayer, integrated in a dual-gate device structure allowing independent control of the electron density and out-of-plane electric field. Under conditions when electron tunneling between the layers is negligible~\cite{pisoni2019absence}, upon electron doping the sample, we observe that the two excitons with opposing dipoles hybridize, displaying unusual behavior distinct from both conventional level crossing and anti-crossing. We show that these observations can be explained by static random coupling between the excitons, which increases with electron density and decreases with temperature. We argue that this phenomenon is indicative of a spatially fluctuating order parameter in the form of interlayer electron coherence, a theoretically predicted many-body state~\cite{zheng1997exchange} that has yet to be unambiguously established experimentally outside of the quantum Hall regime~\cite{sarma2008perspectives,spielman2000resonantly,kellogg2004vanishing,kellogg2002observation,spielman2001observation,fertig1989energy,shi2022bilayer}. Implications of our findings for future experiments and quantum optics applications are discussed.

cond-mat.str-el↗