arXiv subjects

Yuchen Zhou

Publications and source records attributed to Yuchen Zhou.

At least 55 records · Page 3Linked to original sources

arXiv2025

EviNote-RAG: Enhancing RAG Models via Answer-Supportive Evidence Notes

Retrieval-Augmented Generation (RAG) has advanced open-domain question answering by incorporating external information into model reasoning. However, effectively leveraging external information to enhance reasoning presents the following challenges: (1) low signal-to-noise ratio, where answer-supportive external information is diluted by irrelevant material, and (2) error accumulation, which arises in multi-hop reasoning when incomplete or misleading information is incorporated. To address these challenges, we introduce EviNote-RAG, a framework that follows a retrieve-note-answer workflow. Instead of reasoning directly over raw external information, the model first produces Supportive-Evidence Notes (SENs), which concisely preserve answer-critical information and explicitly mark key and uncertainty information to improve accuracy. We further design an entailment-based Evidence Quality Reward (EQR) to ensure that SENs are logically sufficient to derive the final answer, thereby enhancing SENs' quality. Experiments on both in-domain and out-of-domain QA benchmarks show that EviNote-RAG achieves state-of-the-art performance, improving answer accuracy, training stability, robustness, and efficiency. In particular, it yields relative F1 gains of 20% on HotpotQA (+0.093), 40% on Bamboogle (+0.151), and 91% on 2Wiki (+0.256), benefiting from improvements in the reasoning process.

cs.CL↗

arXiv2025

RePainter: Empowering E-commerce Object Removal via Spatial-matting Reinforcement Learning

In web data, product images are central to boosting user engagement and advertising efficacy on e-commerce platforms, yet the intrusive elements such as watermarks and promotional text remain major obstacles to delivering clear and appealing product visuals. Although diffusion-based inpainting methods have advanced, they still face challenges in commercial settings due to unreliable object removal and limited domain-specific adaptation. To tackle these challenges, we propose Repainter, a reinforcement learning framework that integrates spatial-matting trajectory refinement with Group Relative Policy Optimization (GRPO). Our approach modulates attention mechanisms to emphasize background context, generating higher-reward samples and reducing unwanted object insertion. We also introduce a composite reward mechanism that balances global, local, and semantic constraints, effectively reducing visual artifacts and reward hacking. Additionally, we contribute EcomPaint-100K, a high-quality, large-scale e-commerce inpainting dataset, and a standardized benchmark EcomPaint-Bench for fair evaluation. Extensive experiments demonstrate that Repainter significantly outperforms state-of-the-art methods, especially in challenging scenes with intricate compositions. We will release our code and weights upon acceptance.

cs.CV↗

arXiv2025

Logic Unseen: Revealing the Logical Blindspots of Vision-Language Models

Vision-Language Models (VLMs), exemplified by CLIP, have emerged as foundational for multimodal intelligence. However, their capacity for logical understanding remains significantly underexplored, resulting in critical ''logical blindspots'' that limit their reliability in practical applications. To systematically diagnose this, we introduce LogicBench, a comprehensive benchmark with over 50,000 vision-language pairs across 9 logical categories and 4 diverse scenarios: images, videos, anomaly detection, and medical diagnostics. Our evaluation reveals that existing VLMs, even the state-of-the-art ones, fall at over 40 accuracy points below human performance, particularly in challenging tasks like Causality and Conditionality, highlighting their reliance on surface semantics over critical logical structures. To bridge this gap, we propose LogicCLIP, a novel training framework designed to boost VLMs' logical sensitivity through advancements in both data generation and optimization objectives. LogicCLIP utilizes logic-aware data generation and a contrastive learning strategy that combines coarse-grained alignment, a fine-grained multiple-choice objective, and a novel logical structure-aware objective. Extensive experiments demonstrate LogicCLIP's substantial improvements in logical comprehension across all LogicBench domains, significantly outperforming baselines. Moreover, LogicCLIP retains, and often surpasses, competitive performance on general vision-language benchmarks, demonstrating that the enhanced logical understanding does not come at the expense of general alignment. We believe that LogicBench and LogicCLIP will be important resources for advancing VLM logical capabilities.

cs.CV↗

arXiv2025

Faster Diffusion Models via Higher-Order Approximation

In this paper, we explore provable acceleration of diffusion models without any additional retraining. Focusing on the task of approximating a target data distribution in $\mathbb{R}^d$ to within $\varepsilon$ total-variation distance, we propose a principled, training-free sampling algorithm that requires only the order of $$ d^{1+2/K} \varepsilon^{-1/K} $$ score function evaluations (up to log factor) in the presence of accurate scores, where $K>0$ is an arbitrary fixed integer. This result applies to a broad class of target data distributions, without the need for assumptions such as smoothness or log-concavity. Our theory is robust vis-a-vis inexact score estimation, degrading gracefully as the score estimation error increases -- without demanding higher-order smoothness on the score estimates as assumed in previous work. The proposed algorithm draws insight from high-order ODE solvers, leveraging high-order Lagrange interpolation and successive refinement to approximate the integral derived from the probability flow ODE. More broadly, our work develops a theoretical framework towards understanding the efficacy of high-order methods for accelerated sampling.

cs.LG↗

arXiv2025

Oedipus and the Sphinx: Benchmarking and Improving Visual Language Models for Complex Graphic Reasoning

Evaluating the performance of visual language models (VLMs) in graphic reasoning tasks has become an important research topic. However, VLMs still show obvious deficiencies in simulating human-level graphic reasoning capabilities, especially in complex graphic reasoning and abstract problem solving, which are less studied and existing studies only focus on simple graphics. To evaluate the performance of VLMs in complex graphic reasoning, we propose ReasonBench, the first evaluation benchmark focused on structured graphic reasoning tasks, which includes 1,613 questions from real-world intelligence tests. ReasonBench covers reasoning dimensions related to location, attribute, quantity, and multi-element tasks, providing a comprehensive evaluation of the performance of VLMs in spatial, relational, and abstract reasoning capabilities. We benchmark 11 mainstream VLMs (including closed-source and open-source models) and reveal significant limitations of current models. Based on these findings, we propose a dual optimization strategy: Diagrammatic Reasoning Chain (DiaCoT) enhances the interpretability of reasoning by decomposing layers, and ReasonTune enhances the task adaptability of model reasoning through training, all of which improves VLM performance by 33.5\%. All experimental data and code are in the repository: https://huggingface.co/datasets/cistine/ReasonBench.

cs.AI↗

arXiv2025

CSBrain: A Cross-scale Spatiotemporal Brain Foundation Model for EEG Decoding

Understanding and decoding brain activity from electroencephalography (EEG) signals is a fundamental challenge in neuroscience and AI, with applications in cognition, emotion recognition, diagnosis, and brain-computer interfaces. While recent EEG foundation models advance generalized decoding via unified architectures and large-scale pretraining, they adopt a scale-agnostic dense modeling paradigm inherited from NLP and vision. This design neglects a core property of neural activity: cross-scale spatiotemporal structure. EEG task patterns span a wide range of temporal and spatial scales, from short bursts to slow rhythms, and from localized cortical responses to distributed interactions. Ignoring this diversity leads to suboptimal representations and weak generalization. We propose CSBrain, a Cross-scale Spatiotemporal Brain foundation model for generalized EEG decoding. CSBrain introduces: (i) Cross-scale Spatiotemporal Tokenization (CST), which aggregates multi-scale features from localized temporal windows and anatomical brain regions into compact scale-aware tokens; and (ii) Structured Sparse Attention (SSA), which captures cross-window and cross-region dependencies, enhancing scale diversity while removing spurious correlations. CST and SSA are alternately stacked to progressively integrate multi-scale dependencies. Experiments on 11 EEG tasks across 16 datasets show that CSBrain consistently outperforms task-specific and foundation model baselines. These results establish cross-scale modeling as a key inductive bias and position CSBrain as a robust backbone for future brain-AI research.

cs.HC↗

arXiv2025

Where, What, Why: Towards Explainable Driver Attention Prediction

Modeling task-driven attention in driving is a fundamental challenge for both autonomous vehicles and cognitive science. Existing methods primarily predict where drivers look by generating spatial heatmaps, but fail to capture the cognitive motivations behind attention allocation in specific contexts, which limits deeper understanding of attention mechanisms. To bridge this gap, we introduce Explainable Driver Attention Prediction, a novel task paradigm that jointly predicts spatial attention regions (where), parses attended semantics (what), and provides cognitive reasoning for attention allocation (why). To support this, we present W3DA, the first large-scale explainable driver attention dataset. It enriches existing benchmarks with detailed semantic and causal annotations across diverse driving scenarios, including normal conditions, safety-critical situations, and traffic accidents. We further propose LLada, a Large Language model-driven framework for driver attention prediction, which unifies pixel modeling, semantic parsing, and cognitive reasoning within an end-to-end architecture. Extensive experiments demonstrate the effectiveness of LLada, exhibiting robust generalization across datasets and driving conditions. This work serves as a key step toward a deeper understanding of driver attention mechanisms, with significant implications for autonomous driving, intelligent driver training, and human-computer interaction.

cs.CV↗

arXiv2025

Performance Characterizations and Usage Guidelines of Samsung CXL Memory Module Hybrid Prototype

The growing prevalence of data-intensive workloads, such as artificial intelligence (AI), machine learning (ML), high-performance computing (HPC), in-memory databases, and real-time analytics, has exposed limitations in conventional memory technologies like DRAM. While DRAM offers low latency and high throughput, it is constrained by high costs, scalability challenges, and volatility, making it less viable for capacity-bound and persistent applications in modern datacenters. Recently, Compute Express Link (CXL) has emerged as a promising alternative, enabling high-speed, cacheline-granular communication between CPUs and external devices. By leveraging CXL technology, NAND flash can now be used as memory expansion, offering three-fold benefits: byte-addressability, scalable capacity, and persistence at a low cost. Samsung's CXL Memory Module Hybrid (CMM-H) is the first product to deliver these benefits through a hardware-only solution, i.e., it does not incur any OS and IO overheads like conventional block devices. In particular, CMM-H integrates a DRAM cache with NAND flash in a single device to deliver near-DRAM latency. This paper presents the first publicly available study for comprehensive characterizations of an FPGA-based CMM-H prototype. Through this study, we address users' concerns about whether a wide variety of applications can successfully run on a memory device backed by NAND flash medium. Additionally, based on these characterizations, we provide key insights into how to best take advantage of the CMM-H device.

cs.AR↗

arXiv2025

Divergences between Language Models and Human Brains

Do machines and humans process language in similar ways? Recent research has hinted at the affirmative, showing that human neural activity can be effectively predicted using the internal representations of language models (LMs). Although such results are thought to reflect shared computational principles between LMs and human brains, there are also clear differences in how LMs and humans represent and use language. In this work, we systematically explore the divergences between human and machine language processing by examining the differences between LM representations and human brain responses to language as measured by Magnetoencephalography (MEG) across two datasets in which subjects read and listened to narrative stories. Using an LLM-based data-driven approach, we identify two domains that LMs do not capture well: social/emotional intelligence and physical commonsense. We validate these findings with human behavioral experiments and hypothesize that the gap is due to insufficient representations of social/emotional and physical knowledge in LMs. Our results show that fine-tuning LMs on these domains can improve their alignment with human brain responses.

cs.CL↗

arXiv2024

Point-SAM: Promptable 3D Segmentation Model for Point Clouds

The development of 2D foundation models for image segmentation has been significantly advanced by the Segment Anything Model (SAM). However, achieving similar success in 3D models remains a challenge due to issues such as non-unified data formats, poor model scalability, and the scarcity of labeled data with diverse masks. To this end, we propose a 3D promptable segmentation model Point-SAM, focusing on point clouds. We employ an efficient transformer-based architecture tailored for point clouds, extending SAM to the 3D domain. We then distill the rich knowledge from 2D SAM for Point-SAM training by introducing a data engine to generate part-level and object-level pseudo-labels at scale from 2D SAM. Our model outperforms state-of-the-art 3D segmentation models on several indoor and outdoor benchmarks and demonstrates a variety of applications, such as interactive 3D annotation and zero-shot 3D instance proposal. Codes and demo can be found at https://github.com/zyc00/Point-SAM.

cs.CV↗

arXiv2024

Probing Charge Dynamics in Amorphous Oxide Semiconductors by Time-of-flight Microwave Impedance Microscopy

The unique electronic properties of amorphous indium gallium zinc oxide (a-IGZO) thin films are closely associated with the complex charge dynamics of the materials. Conventional studies of charge transport in a-IGZO usually involve steady-state or transient measurements on field-effect transistors. Here, we employed microwave impedance microscopy to carry out position-dependent time-of-flight (TOF) experiments on a-IGZO devices, which offer spatial and temporal information of the underlying transport dynamics. The drift mobility calculated from the delay time between carrier injection and onset of TOF response is 2 - 3 cm2/Vs, consistent with the field-effect mobility from device measurements. The spatiotemporal conductivity data can be nicely fitted to a two-step function, corresponding to two coexisting mechanisms with a typical timescale of milliseconds. The competition between multiple-trap-and-release conduction through band-tail states and hopping conduction through deep trap states is evident from the fitting parameters. The underlying length scale and time scale of charge dynamics in a-IGZO are of fundamental importance for transparent and flexible nanoelectronics and optoelectronics, as well as emerging back-end-of-line applications.

cond-mat.mtrl-sci↗

arXiv2024

Intelligent Reflecting Surface-Assisted Symbiotic Radio Systems: A Double-Reflection Covert Communication Design

We investigate covert communication in an intelligent reflecting surface (IRS)-assisted symbiotic radio (SR) system under the parasitic SR (PSR) and the commensal SR (CSR) cases, where an IRS is exploited to create a double reflection link for legitimate users and degrade the detection performance of the warden (W). Specifically, we derive an analytical expression for the average detection error probability of W and design an optimal strategy to determine the transmit power and backscatter reflection coefficient. To further enhance the covert performance, the joint optimization of the source transmit power, backscatter device (BD) reflection coefficient, and IRS phase-shifter is formulated as an expectation-based quadratic-fractional (EQF) problem. By reformulating the original problem into a fraction-eliminated backscatter power leakage minimization problem, we further develop the phase alignment pursuit and the power leakage minimization algorithms for the PSR and the CSR cases, respectively. Numerical results confirm the accuracy of the derived results and the superiority of our proposed strategy in terms of covertness.

cs.IT↗

arXiv2024

Deflated HeteroPCA: Overcoming the curse of ill-conditioning in heteroskedastic PCA

This paper is concerned with estimating the column subspace of a low-rank matrix $\boldsymbol{X}^\star \in \mathbb{R}^{n_1\times n_2}$ from contaminated data. How to obtain optimal statistical accuracy while accommodating the widest range of signal-to-noise ratios (SNRs) becomes particularly challenging in the presence of heteroskedastic noise and unbalanced dimensionality (i.e., $n_2\gg n_1$). While the state-of-the-art algorithm $\textsf{HeteroPCA}$ emerges as a powerful solution for solving this problem, it suffers from "the curse of ill-conditioning," namely, its performance degrades as the condition number of $\boldsymbol{X}^\star$ grows. In order to overcome this critical issue without compromising the range of allowable SNRs, we propose a novel algorithm, called $\textsf{Deflated-HeteroPCA}$, that achieves near-optimal and condition-number-free theoretical guarantees in terms of both $\ell_2$ and $\ell_{2,\infty}$ statistical accuracy. The proposed algorithm divides the spectrum of $\boldsymbol{X}^\star$ into well-conditioned and mutually well-separated subblocks, and applies $\textsf{HeteroPCA}$ to conquer each subblock successively. Further, an application of our algorithm and theory to two canonical examples -- the factor model and tensor PCA -- leads to remarkable improvement for each application.

math.ST↗

arXiv2024

High Mobility SiGe/Ge 2DHG Heterostructure Quantum Wells for Semiconductor Hole Spin Qubits

Strong spin-orbit coupling and relatively weak hyperfine interactions make germanium hole spin qubits a promising candidate for semiconductor quantum processors. The two-dimensional hole gas structure of strained Ge quantum wells serves as the primary material platform for spin hole qubits.A low disorder material environment is essential for this process. In this work, we fabricated a Ge/SiGe heterojunction with a 60 nm buried quantum well layer on a Si substrate using reduced pressure chemical vapor deposition technology. At a temperature of 16 mK, when the carrier density is 1.87*10^11/cm2, we obtained a mobility as high as 308.64*10^4cm2/Vs. Concurrently, double quantum dot and planar germanium coupling with microwave cavities were also successfully achieved.This fully demonstrates that this structure can be used for the preparation of higher-performance hole spin qubits.

cond-mat.mes-hall↗

arXiv2024

Self-Supervised Contrastive Graph Clustering Network via Structural Information Fusion

Graph clustering, a classical task in graph learning, involves partitioning the nodes of a graph into distinct clusters. This task has applications in various real-world scenarios, such as anomaly detection, social network analysis, and community discovery. Current graph clustering methods commonly rely on module pre-training to obtain a reliable prior distribution for the model, which is then used as the optimization objective. However, these methods often overlook deeper supervised signals, leading to sub-optimal reliability of the prior distribution. To address this issue, we propose a novel deep graph clustering method called CGCN. Our approach introduces contrastive signals and deep structural information into the pre-training process. Specifically, CGCN utilizes a contrastive learning mechanism to foster information interoperability among multiple modules and allows the model to adaptively adjust the degree of information aggregation for different order structures. Our CGCN method has been experimentally validated on multiple real-world graph datasets, showcasing its ability to boost the dependability of prior clustering distributions acquired through pre-training. As a result, we observed notable enhancements in the performance of the model.

cs.LG↗

arXiv2024

Learning from Observer Gaze:Zero-Shot Attention Prediction Oriented by Human-Object Interaction Recognition

Most existing attention prediction research focuses on salient instances like humans and objects. However, the more complex interaction-oriented attention, arising from the comprehension of interactions between instances by human observers, remains largely unexplored. This is equally crucial for advancing human-machine interaction and human-centered artificial intelligence. To bridge this gap, we first collect a novel gaze fixation dataset named IG, comprising 530,000 fixation points across 740 diverse interaction categories, capturing visual attention during human observers cognitive processes of interactions. Subsequently, we introduce the zero-shot interaction-oriented attention prediction task ZeroIA, which challenges models to predict visual cues for interactions not encountered during training. Thirdly, we present the Interactive Attention model IA, designed to emulate human observers cognitive processes to tackle the ZeroIA problem. Extensive experiments demonstrate that the proposed IA outperforms other state-of-the-art approaches in both ZeroIA and fully supervised settings. Lastly, we endeavor to apply interaction-oriented attention to the interaction recognition task itself. Further experimental results demonstrate the promising potential to enhance the performance and interpretability of existing state-of-the-art HOI models by incorporating real human attention data from IG and attention labels generated by IA.

cs.CV↗

arXiv2024

Advancing Multimodal Medical Capabilities of Gemini

Many clinical tasks require an understanding of specialized data, such as medical images and genomics, which is not typically found in general-purpose large multimodal models. Building upon Gemini's multimodal models, we develop several models within the new Med-Gemini family that inherit core capabilities of Gemini and are optimized for medical use via fine-tuning with 2D and 3D radiology, histopathology, ophthalmology, dermatology and genomic data. Med-Gemini-2D sets a new standard for AI-based chest X-ray (CXR) report generation based on expert evaluation, exceeding previous best results across two separate datasets by an absolute margin of 1% and 12%, where 57% and 96% of AI reports on normal cases, and 43% and 65% on abnormal cases, are evaluated as "equivalent or better" than the original radiologists' reports. We demonstrate the first ever large multimodal model-based report generation for 3D computed tomography (CT) volumes using Med-Gemini-3D, with 53% of AI reports considered clinically acceptable, although additional research is needed to meet expert radiologist reporting quality. Beyond report generation, Med-Gemini-2D surpasses the previous best performance in CXR visual question answering (VQA) and performs well in CXR classification and radiology VQA, exceeding SoTA or baselines on 17 of 20 tasks. In histopathology, ophthalmology, and dermatology image classification, Med-Gemini-2D surpasses baselines across 18 out of 20 tasks and approaches task-specific model performance. Beyond imaging, Med-Gemini-Polygenic outperforms the standard linear polygenic risk score-based approach for disease risk prediction and generalizes to genetically correlated diseases for which it has never been trained. Although further development and evaluation are necessary in the safety-critical medical domain, our results highlight the potential of Med-Gemini across a wide range of medical tasks.

cs.CV↗

arXiv2024

CEBin: A Cost-Effective Framework for Large-Scale Binary Code Similarity Detection

Binary code similarity detection (BCSD) is a fundamental technique for various application. Many BCSD solutions have been proposed recently, which mostly are embedding-based, but have shown limited accuracy and efficiency especially when the volume of target binaries to search is large. To address this issue, we propose a cost-effective BCSD framework, CEBin, which fuses embedding-based and comparison-based approaches to significantly improve accuracy while minimizing overheads. Specifically, CEBin utilizes a refined embedding-based approach to extract features of target code, which efficiently narrows down the scope of candidate similar code and boosts performance. Then, it utilizes a comparison-based approach that performs a pairwise comparison on the candidates to capture more nuanced and complex relationships, which greatly improves the accuracy of similarity detection. By bridging the gap between embedding-based and comparison-based approaches, CEBin is able to provide an effective and efficient solution for detecting similar code (including vulnerable ones) in large-scale software ecosystems. Experimental results on three well-known datasets demonstrate the superiority of CEBin over existing state-of-the-art (SOTA) baselines. To further evaluate the usefulness of BCSD in real world, we construct a large-scale benchmark of vulnerability, offering the first precise evaluation scheme to assess BCSD methods for the 1-day vulnerability detection task. CEBin could identify the similar function from millions of candidate functions in just a few seconds and achieves an impressive recall rate of $85.46\%$ on this more practical but challenging task, which are several order of magnitudes faster and $4.07\times$ better than the best SOTA baseline. Our code is available at https://github.com/Hustcw/CEBin.

cs.SE↗