arXiv subjects

Yang Hu

Publications and source records attributed to Yang Hu.

At least 91 records · Page 5Linked to original sources

arXiv2025

MoBiLE: Efficient Mixture-of-Experts Inference on Consumer GPU with Mixture of Big Little Experts

Mixture-of-Experts (MoE) models have recently demonstrated exceptional performance across a diverse range of applications. The principle of sparse activation in MoE models facilitates an offloading strategy, wherein active experts are maintained in GPU HBM, while inactive experts are stored in CPU DRAM. The efficacy of this approach, however, is fundamentally constrained by the limited bandwidth of the CPU-GPU interconnect. To mitigate this bottleneck, existing approaches have employed prefetching to accelerate MoE inference. These methods attempt to predict and prefetch the required experts using specially trained modules. Nevertheless, such techniques are often encumbered by significant training overhead and have shown diminished effectiveness on recent MoE models with fine-grained expert segmentation. In this paper, we propose MoBiLE, a plug-and-play offloading-based MoE inference framework with \textit{mixture of big-little experts}. It reduces the number of experts for unimportant tokens to half for acceleration while maintaining full experts for important tokens to guarantee model quality. Further, a dedicated fallback and prefetching mechanism is designed for switching between little and big experts to improve memory efficiency. We evaluate MoBiLE on four typical modern MoE architectures and challenging generative tasks. Our results show that MoBiLE achieves a speedup of 1.60x to 1.72x compared to the baseline on a consumer GPU system, with negligible degradation in accuracy.

cs.CL↗

arXiv2025

GoalFlow: Goal-Driven Flow Matching for Multimodal Trajectories Generation in End-to-End Autonomous Driving

We propose GoalFlow, an end-to-end autonomous driving method for generating high-quality multimodal trajectories. In autonomous driving scenarios, there is rarely a single suitable trajectory. Recent methods have increasingly focused on modeling multimodal trajectory distributions. However, they suffer from trajectory selection complexity and reduced trajectory quality due to high trajectory divergence and inconsistencies between guidance and scene information. To address these issues, we introduce GoalFlow, a novel method that effectively constrains the generative process to produce high-quality, multimodal trajectories. To resolve the trajectory divergence problem inherent in diffusion-based methods, GoalFlow constrains the generated trajectories by introducing a goal point. GoalFlow establishes a novel scoring mechanism that selects the most appropriate goal point from the candidate points based on scene information. Furthermore, GoalFlow employs an efficient generative method, Flow Matching, to generate multimodal trajectories, and incorporates a refined scoring mechanism to select the optimal trajectory from the candidates. Our experimental results, validated on the Navsim\cite{Dauner2024_navsim}, demonstrate that GoalFlow achieves state-of-the-art performance, delivering robust multimodal trajectories for autonomous driving. GoalFlow achieved PDMS of 90.3, significantly surpassing other methods. Compared with other diffusion-policy-based methods, our approach requires only a single denoising step to obtain excellent performance. The code is available at https://github.com/YvanYin/GoalFlow.

cs.CV↗

arXiv2025

AdaFusion: Prompt-Guided Inference with Adaptive Fusion of Pathology Foundation Models

Pathology foundation models (PFMs) have demonstrated strong representational capabilities through self-supervised pre-training on large-scale, unannotated histopathology image datasets. However, their diverse yet opaque pretraining contexts, shaped by both data-related and structural/training factors, introduce latent biases that hinder generalisability and transparency in downstream applications. In this paper, we propose AdaFusion, a novel prompt-guided inference framework that, to our knowledge, is among the very first to dynamically integrate complementary knowledge from multiple PFMs. Our method compresses and aligns tile-level features from diverse models and employs a lightweight attention mechanism to adaptively fuse them based on tissue phenotype context. We evaluate AdaFusion on three real-world benchmarks spanning treatment response prediction, tumour grading, and spatial gene expression inference. Our approach consistently surpasses individual PFMs across both classification and regression tasks, while offering interpretable insights into each model's biosemantic specialisation. These results highlight AdaFusion's ability to bridge heterogeneous PFMs, achieving both enhanced performance and interpretability of model-specific inductive biases.

cs.CV↗

arXiv2025

MCBP: A Memory-Compute Efficient LLM Inference Accelerator Leveraging Bit-Slice-enabled Sparsity and Repetitiveness

Large language models (LLMs) face significant inference latency due to inefficiencies in GEMM operations, weight access, and KV cache access, especially in real-time scenarios. This highlights the need for a versatile compute-memory efficient accelerator. Unfortunately, existing Transformer accelerators struggle to address both aspects simultaneously, as they focus on value-level processing, missing fine-grained opportunities to optimize computation and memory collaboratively. This paper introduces MCBP, a bit-grained compute-memory efficient algorithm-hardware co-design that leverages bit-slice (BS) enabled repetitiveness and sparsity to accelerate LLM inference. MCBP features three key innovations: 1) BS-repetitiveness-enabled computation reduction (BRCR), which eliminates redundant GEMM computations via leveraging redundancy hidden among BS vectors; 2) BS-sparsity-enabled two-state coding (BSTC), which reduces weight access via exploiting significant sparsity in high-order bit-slice weight; 3) Bit-grained progressive prediction (BGPP), which reduces KV cache access by leveraging early-termination-based bit-grained prediction. These techniques, supported by custom accelerator designs, effectively alleviate the burden in GEMM, weight access, and KV cache access. Extensive experiments on 26 benchmarks show that MCBP achieves 9.43x speed up and 31.1x higher energy efficiency than Nvidia A100 GPU. Compared to SOTA Transformer accelerators, MCBP achieves 35x, 5.2x and 3.2x energy saving than Spatten, FACT and SOFA, respectively.

cs.AR↗

arXiv2025

Quartz phononic crystal resonators for hybrid acoustic quantum memories

Circuit quantum acoustodynamics systems have emerged as a promising platform for quantum information by coupling superconducting qubits to mechanical resonators, with their long-lived mechanical modes serving as quantum memories. We demonstrate suspended quartz phononic crystal resonators at 100 MHz with millisecond lifetimes at 8 K. With a contactless electrode geometry suppressing both two-level system losses and other electrode-induced energy dissipation, we evaluate the piezoelectric coupling rate between the mechanical modes and fluxonium qubits (resonant coupling) and transmon qubits (parametric coupling mediated by a Josephon-junction-based three-wave mixer). We further discuss multi-period defect geometries for enhancing these coupling rates.

quant-ph↗

arXiv2025

Ultrathin oxide freestanding membranes with large-scale continuity and structural perfection

Freestanding oxide membranes offer integration with advanced semiconductor platforms, unlocking opportunities for flexible electronics, silicon-based spintronics, neuromorphic computing, and high-performance energy technologies. Scalable fabrication of such membranes is essential for bridging fundamental discoveries in complex oxides with practical device deployment. However, the lateral dimensions of crack- and wrinkle-free membranes remain limited to millimeter scales, forming a critical bottleneck for large-area architectures. Overcoming this challenge demands strategies that preserve crystalline quality while suppressing defect transfer during release. Here, we demonstrate an approach based on a water-soluble sacrificial layer of super-tetragonal Sr4Al2O7, enabling the fabrication of ultrathin, crack-free, and wrinkle-free free-standing oxide membranes spanning centimeter-scale areas. This method is broadly applicable to a wide range of oxides and establishes a new pathway toward large-scale silicon integration and flexible oxide technologies. Nevertheless, dissolution of the sacrificial layer introduces oxygen vacancies into the SrRuO3 membranes, with diffusion depths reaching six unit cells, leading to anomalous "up-and-down" transport behavior. Although post-annealing can eliminate these vacancies, the required temperatures are incompatible with CMOS processes. Therefore, ultrathin freestanding membranes fabricated by water-assisted lift-off still face critical challenges for integration into miniaturized silicon-based oxide devices.

cond-mat.mtrl-sci↗

arXiv2025

SN 2021aaev: a Hydrogen-Rich Superluminous Supernova with Early Flash and Long-Lived Circumstellar Interaction in an Unusual Host Environment

We present photometric and spectroscopic observations of SN\,2021aaev, a hydrogen-rich, superluminous supernova with persistent (at least $\sim100$ days) narrow Balmer lines (SLSN-IIn) at redshift $z=0.1557$. We observed SN\,2021aaev to rise in $32.5 \pm 1.0$ days since first light and reach a peak absolute magnitude of $-21.46 \pm 0.01$ in the ATLAS $o$ band. The pre-peak spectra resemble those of typical SNe IIn with flash-ionization features arising from the interaction with a dense, confined circumstellar medium (CSM), albeit the flash timescale is longer than usual ($>20$ days). Post peak, the narrow emission lines evolve slowly, and the absence of ejecta features indicates strong deceleration by the CSM. The total radiated energy (about $1.41\times10^{51}$~ergs) is possible with a low-mass (1--$2\,M_{\odot}$) ejecta ploughing into a massive (9--$19\,M_{\odot}$), extended (outer radius $>1\times10^{16}$~cm) H-rich CSM, or alternatively, with magnetar-powered models. Interestingly, the host environment consists of a spiral galaxy with a red substructure in the south-eastern part, and the SN's exact location coincided with the quiescent red substructure (star-formation rate$=0.02^{+0.13}_{-0.02}\,M_{\odot}$~yr$^{-1}$). Given the atypical environment and the obscuring effect of the massive CSM, a thermonuclear (Type Ia-CSM) origin cannot be ruled out. Altogether, SN\,2021aaev is a compelling case to study the diversity of SLSN-IIn features and their host environment.

astro-ph.HE↗

arXiv2025

Pre-trained Audio Transformer as a Foundational AI Tool for Gravitational Waves

As gravitational wave detectors become more advanced and sensitive, the number of signals recorded by Advanced LIGO and Virgo from merging compact objects is expected to rise dramatically. This surge in detection rates necessitates the development of adaptable, scalable, and efficient tools capable of addressing a wide range of tasks in gravitational wave astronomy. Foundational AI models present a transformative opportunity in this context by providing a unified framework that can be fine tuned for diverse applications while leveraging the power of large scale pre training. In this work, we explore how advanced transformer models, specifically Whisper by OpenAI, can be adapted as a foundational model for gravitational wave data analysis. By fine tuning the encoder model of Whisper, originally trained on extensive audio data, and combining it with neural networks for specialized tasks, we achieve reliable results in detecting astrophysical signals and classifying transient noise artifacts or glitches. This represents the first application of open source transformer models, pre trained on unrelated tasks, for gravitational wave research, demonstrating their potential to enable versatile and efficient data analysis in the era of rapidly increasing detection rates.

gr-qc↗

arXiv2025

HPCTransCompile: An AI Compiler Generated Dataset for High-Performance CUDA Transpilation and LLM Preliminary Exploration

The rapid growth of deep learning has driven exponential increases in model parameters and computational demands. NVIDIA GPUs and their CUDA-based software ecosystem provide robust support for parallel computing, significantly alleviating computational bottlenecks. Meanwhile, due to the cultivation of user programming habits and the high performance of GPUs, the CUDA ecosystem has established a dominant position in the field of parallel software. This dominance requires other hardware platforms to support CUDA-based software with performance portability. However, translating CUDA code to other platforms poses significant challenges due to differences in parallel programming paradigms and hardware architectures. Existing approaches rely on language extensions, domain-specific languages (DSLs), or compilers but face limitations in workload coverage and generalizability. Moreover, these methods often incur substantial development costs. Recently, LLMs have demonstrated extraordinary potential in various vertical domains, especially in code-related tasks. However, the performance of existing LLMs in CUDA transpilation, particularly for high-performance code, remains suboptimal. To address these challenges, we propose a novel framework for generating high-performance CUDA and corresponding platform code pairs, leveraging AI compiler and automatic optimization technology. We further enhance the framework with a graph-based data augmentation method and introduce HPCTransEval, a benchmark for evaluating LLM performance on CUDA transpilation. We conduct experiments using CUDA-to-CPU transpilation as a case study on leading LLMs. The speedup ratio of the CPU operators has an average improvemnet of 43.8\%, highlighting the potential of LLMs to address compatibility challenges within the CUDA ecosystem. Our code is available at https://github.com/PJLAB-CHIP/HPCTransCompile.

cs.DC↗

arXiv2025

Automatic Phase Calibration for High-resolution mmWave Sensing via Ambient Radio Anchors

Millimeter-wave (mmWave) radar systems with large array have pushed radar sensing into a new era, thanks to their high angular resolution. However, our long-term experiments indicate that array elements exhibit phase drift over time and require periodic phase calibration to maintain high-resolution, creating an obstacle for practical high-resolution mmWave sensing. Unfortunately, existing calibration methods are inadequate for periodic recalibration, either because they rely on artificial references or fail to provide sufficient precision. To address this challenge, we introduce AutoCalib, the first framework designed to automatically and accurately calibrate high-resolution mmWave radars by identifying Ambient Radio Anchors (ARAs)-naturally existing objects in ambient environments that offer stable phase references. AutoCalib achieves calibration by first generating spatial spectrum templates based on theoretical electromagnetic characteristics. It then employs a pattern-matching and scoring mechanism to accurately detect these anchors and select the optimal one for calibration. Extensive experiments across 11 environments demonstrate that AutoCalib capable of identifying ARAs that existing methods miss due to their focus on strong reflectors. AutoCalib's calibration performance approaches corner reflectors (74% phase error reduction) while outperforming existing methods by 83%. Beyond radar calibration, AutoCalib effectively supports other phase-dependent applications like handheld imaging, delivering 96% of corner reflector calibration performance without artificial references.

eess.SP↗

arXiv2025

RS-Coded Adaptive Dynamic Network for Reliable Long-Term Information Transmission in Disturbed Multimode Fiber

Multimode fiber (MMF), due to its large core diameter and high mode capacity, holds potential in high-speed communications. However, inherent modal dispersion causes output speckle distortion, and transmission characteristics are sensitive to environmental disturbances, limiting its reliable application. Conventional transmission matrix (TM) methods face challenges such as complex calibration and environmental sensitivity. Although current deep learning approaches demonstrate reconstruction potential, they struggle to overcome error accumulation caused by fiber mode drift and lack sufficient environmental adaptability. To address this, the present study proposes an adaptive transmission framework named Residual Reed-Solomon Dynamic Network (RRSDN), which integrates Reed-Solomon (RS) error correction coding with deep residual learning forming a closed-loop system that jointly optimizes encoding, transmission, and reconstruction, to tackle the key challenges of mode instability and error accumulation in dynamic scattering channels. Experimentally, high-fidelity real-time transmission of a 16*16 pixel video stream (H.265 compressed) with zero frame loss and 100% symbol accuracy was achieved under conditions of a 100-meter MMF with manually applied disturbances and no temperature control. This work proposes a solution for stable optical transmission in complex channels. Plus, it integrates error correction coding with neural network training, laying the foundation for adaptive optical systems in longer-distance and more complex scenarios.

physics.optics↗

arXiv2025

A Study on Individual Spatiotemporal Activity Generation Method Using MCP-Enhanced Chain-of-Thought Large Language Models

Human spatiotemporal behavior simulation is critical for urban planning research, yet traditional rule-based and statistical approaches suffer from high computational costs, limited generalizability, and poor scalability. While large language models (LLMs) show promise as "world simulators," they face challenges in spatiotemporal reasoning including limited spatial cognition, lack of physical constraint understanding, and group homogenization tendencies. This paper introduces a framework integrating chain-of-thought (CoT) reasoning with Model Context Protocol (MCP) to enhance LLMs' capability in simulating spatiotemporal behaviors that correspond with validation data patterns. The methodology combines human-like progressive reasoning through a five-stage cognitive framework with comprehensive data processing via six specialized MCP tool categories: temporal management, spatial navigation, environmental perception, personal memory, social collaboration, and experience evaluation. Experiments in Shanghai's Lujiazui district validate the framework's effectiveness across 1,000 generated samples. Results demonstrate high similarity with real mobile signaling data, achieving generation quality scores of 7.86 to 8.36 across different base models. Parallel processing experiments show efficiency improvements, with generation times decreasing from 1.30 to 0.17 minutes per sample when scaling from 2 to 12 processes. This work contributes to integrating CoT reasoning with MCP for urban behavior modeling, advancing LLMs applications in urban computing and providing a practical approach for synthetic mobility data generation. The framework offers a foundation for smart city planning, transportation forecasting, and participatory urban design applications.

cs.AI↗

arXiv2025

SLOT: Sample-specific Language Model Optimization at Test-time

We propose SLOT (Sample-specific Language Model Optimization at Test-time), a novel and parameter-efficient test-time inference approach that enhances a language model's ability to more accurately respond to individual prompts. Existing Large Language Models (LLMs) often struggle with complex instructions, leading to poor performances on those not well represented among general samples. To address this, SLOT conducts few optimization steps at test-time to update a light-weight sample-specific parameter vector. It is added to the final hidden layer before the output head, and enables efficient adaptation by caching the last layer features during per-sample optimization. By minimizing the cross-entropy loss on the input prompt only, SLOT helps the model better aligned with and follow each given instruction. In experiments, we demonstrate that our method outperforms the compared models across multiple benchmarks and LLMs. For example, Qwen2.5-7B with SLOT achieves an accuracy gain of 8.6% on GSM8K from 57.54% to 66.19%, while DeepSeek-R1-Distill-Llama-70B with SLOT achieves a SOTA accuracy of 68.69% on GPQA among 70B-level models. Our code is available at https://github.com/maple-research-lab/SLOT.

cs.CL↗

arXiv2025

Beginning with You: Perceptual-Initialization Improves Vision-Language Representation and Alignment

We introduce Perceptual-Initialization (PI), a paradigm shift in visual representation learning that incorporates human perceptual structure during the initialization phase rather than as a downstream fine-tuning step. By integrating human-derived triplet embeddings from the NIGHTS dataset to initialize a CLIP vision encoder, followed by self-supervised learning on YFCC15M, our approach demonstrates significant zero-shot performance improvements, without any task-specific fine-tuning, across 29 zero shot classification and 2 retrieval benchmarks. On ImageNet-1K, zero-shot gains emerge after approximately 15 epochs of pretraining. Benefits are observed across datasets of various scales, with improvements manifesting at different stages of the pretraining process depending on dataset characteristics. Our approach consistently enhances zero-shot top-1 accuracy, top-5 accuracy, and retrieval recall (e.g., R@1, R@5) across these diverse evaluation tasks, without requiring any adaptation to target domains. These findings challenge the conventional wisdom of using human-perceptual data primarily for fine-tuning and demonstrate that embedding human perceptual structure during early representation learning yields more capable and vision-language aligned systems that generalize immediately to unseen tasks. Our work shows that "beginning with you", starting with human perception, provides a stronger foundation for general-purpose vision-language intelligence.

cs.CV↗

arXiv2025

Shifting Attention to You: Personalized Brain-Inspired AI Models

The integration of human and artificial intelligence offers a powerful avenue for advancing our understanding of information processing, as each system provides unique computational insights. However, despite the promise of human-AI integration, current AI models are largely trained on massive datasets, optimized for population-level performance, lacking mechanisms to align their computations with individual users' perceptual semantics and neural dynamics. Here we show that integrating human behavioral insights and millisecond scale neural data within a fine tuned CLIP based model not only captures generalized and individualized aspects of perception but also over doubles behavioral performance compared to the unmodified CLIP baseline. By embedding human inductive biases and mirroring dynamic neural processes during training, personalized neural fine tuning improves predictions of human similarity judgments and tracks the temporal evolution of individual neural responses. Our work establishes a novel, interpretable framework for designing adaptive AI systems, with broad implications for neuroscience, personalized medicine, and human-computer interaction.

q-bio.NC↗

arXiv2025

The Schur complements for $SDD_{1}$ matrices and their application to linear complementarity problems

In this paper we propose a new scaling method to study the Schur complements of $SDD_{1}$ matrices. Its core is related to the non-negative property of the inverse $M$-matrix, while numerically improving the Quotient formula. Based on the Schur complement and a novel norm splitting manner, we establish an upper bound for the infinity norm of the inverse of $SDD_{1}$ matrices, which depends solely on the original matrix entries. We apply the new bound to derive an error bound for linear complementarity problems of $B_{1}$-matrices. Additionally, new lower and upper bounds for the determinant of $SDD_{1}$ matrices are presented. Numerical experiments validate the effectiveness and superiority of our results.

math.NA↗

arXiv2025

A Model-Based Approach to Imitation Learning through Multi-Step Predictions

Imitation learning is a widely used approach for training agents to replicate expert behavior in complex decision-making tasks. However, existing methods often struggle with compounding errors and limited generalization, due to the inherent challenge of error correction and the distribution shift between training and deployment. In this paper, we present a novel model-based imitation learning framework inspired by model predictive control, which addresses these limitations by integrating predictive modeling through multi-step state predictions. Our method outperforms traditional behavior cloning numerical benchmarks, demonstrating superior robustness to distribution shift and measurement noise both in available data and during execution. Furthermore, we provide theoretical guarantees on the sample complexity and error bounds of our method, offering insights into its convergence properties.

cs.LG↗

arXiv2025

Underwater Robotic Simulators Review for Autonomous System Development

The increasing complexity of underwater robotic systems has led to a surge in simulation platforms designed to support perception, planning, and control tasks in marine environments. However, selecting the most appropriate underwater robotic simulator (URS) remains a challenge due to wide variations in fidelity, extensibility, and task suitability. This paper presents a comprehensive review and comparative analysis of five state-of-the-art, ROS-compatible, open-source URSs: Stonefish, DAVE, HoloOcean, MARUS, and UNav-Sim. Each simulator is evaluated across multiple criteria including sensor fidelity, environmental realism, sim-to-real capabilities, and research impact. We evaluate them across architectural design, sensor and physics modeling, task capabilities, and research impact. Additionally, we discuss ongoing challenges in sim-to-real transfer and highlight the need for standardization and benchmarking in the field. Our findings aim to guide practitioners in selecting effective simulation environments and inform future development of more robust and transferable URSs.

cs.RO↗