arXiv subjects

Yichen Zhao

Publications and source records attributed to Yichen Zhao.

17 recordsLinked to original sources

arXiv2026

HAT-4D: Lifting Monocular Video for 4D Multi-Object Interactions via Human-Agent Collaboration

Extracting dynamic 4D object interactions from massive, in-the-wild monocular videos offers a highly efficient data collection pathway for scaling Embodied AI and training VLAs. However, existing monocular 4D reconstruction methods primarily focus on isolated objects, often failing under the severe occlusions and complex dynamics inherent in multi-object interactions. To bridge this gap, we propose HAT-4D, the first agentic framework designed to reconstruct the 3D geometry, temporal dynamics, and physical interactions of multiple objects from a single video. By integrating VLMs with a multi-level human-in-the-loop feedback mechanism, HAT-4D efficiently resolves depth ambiguities and interaction-induced occlusions during 3D generation and 4D propagation, yielding physically plausible assets without relying on expensive multicamera rigs. As a scalable data engine, HAT-4D facilitates the creation of MVOIK-4D, an open-world benchmark for monocular 4D interaction reconstruction, accompanied by a novel multi-dimensional evaluation protocol focused on physical plausibility and temporal consistency. Extensive experiments demonstrate that HAT-4D achieves SOTA performance on most evaluation metrics, while maintaining competitive semantic alignment. Ablation studies show that introducing a small amount of human feedback improves interaction reconstruction. Moreover, the data produced by HAT-4D effectively improves baseline performance when used for fine-tuning. Our data and code are available at https://lijiaxin0111.github.io/HAT4D/

cs.CV↗

arXiv2026

VHDLSuite: Unified Pipeline for LLM VHDL Generation with Data Synthesis and Evaluation

Large Language Models (LLM) have shown impressive capabilities in Register Transfer Level (RTL) code generation, particularly for Verilog. However, evaluating their performance with other Hardware Description Languages (HDL), especially VHDL, remains limited although its distinct language characteristics, such as stricter semantic rules, introduce evaluation considerations that differ from Verilog. This lack of coverage restricts fully understanding of how well current models generalize across hardware design languages with differing structures and semantics. To address this gap, we introduce VHDLSuite, a benchmark-centered infrastructure for scalable VHDL generation evaluation, integrating automated benchmark synthesis, executable validation, and multi-model diagnostic analysis. First, we propose a data pipeline that automatically converts Verilog designs and their accompanying testbenches into executable VHDL benchmark instances, followed by VUnit/GHDL-based validation to ensure each released task is compilable, runnable, and consistently checkable in the VHDL environment. Second, we introduce VHDLBench, a benchmark with over 200 VHDL problems with complete and validated testbenches across a wide range of complexity levels. Third, we extensively evaluate cutting-edge LLMs and uncover key challenges specific on LLM-aided VHDL generation. Our findings provide important insights and support future work in multi-language hardware design automation.Our data pipeline, benchmark, and evaluation framework will be open-sourced.

cs.AR↗

arXiv2026

KISS - Knowledge Infrastructure for Scientific Simulation: A Scaffolding for Agentic Earth Science

Process-based simulation models encode decades of scientific understanding across the Earth sciences, yet the communities most exposed to climate risk and resource scarcity are the least able to use them. Here, we introduce knowledge infrastructure (KI), an agent-actionable scaffold that externalizes expertise into validated modelling operators, staged domain protocols, and diagnostic recovery mechanisms. Across a 3,000-trial coupled-hydrology benchmark, agents equipped with KI produced physically plausible, verifiable end-to-end simulations in up to 84% of trials, while agents without KI plateaued below 40%. KI generalizes across disciplines. We packaged its construction into a Knowledge Dissection Toolkit (KDT) that autonomously produced KI enabling end-to-end agent execution of 117 additional process-based models across 14 Earth-science domains. Across all 119 KIs, modelling decisions and failure remedies converged despite different underlying physics, showing that operational expertise is structured and extractable rather than ad hoc. Demonstrations show KI-equipped agents lowering both the access barrier between non-specialist users and process-based simulation, and the integration barrier between modelling communities. Through this scaffold, process-based science can then evolve as a living scientific commons, answerable to whoever needs to know and extendable by whoever can contribute.

cs.AI↗

arXiv2026

STAR-Filter: Efficient Convex Free-Space Approximation via Starshaped Set Filtering in Noisy Environments

Approximating collision-free space is fundamental to robot planning in complex environments. Convex geometric representations, such as polytopes and ellipsoids, are widely employed due to their structural properties, which can be easily integrated with convex optimization. Iterative optimization-based inflation methods can generate large volume polytopes in cluttered environments, but their efficiency degrades as the obstacle set becomes more complex or when sensor data are noisy. These methods are also sensitive to initialization and often rely on accurate geometric models. In this paper, we propose the STAR-Filter, a lightweight framework that employs starshaped set construction as a fast filter for convex region generation in collision-free space. By identifying obstacle points as active supporting constraints, the proposed method significantly reduces redundant computation while preserving feasibility and robustness to sensor noise. We provide theoretical and numerical analyses that characterize the structural properties of the starshaped set and proposed pipeline in environments of varying complexity. Simulation results show that the proposed framework achieves the lowest computation time and reduces conservativeness in polytope generation for real-world noisy and large-scale data. We demonstrate the effectiveness of the framework for Safe Flight Corridor (SFC) generation and agile quadrotor planning in noisy environments.

cs.RO↗

arXiv2026

MMRad-22K: A Structured Multimodal Evidence Dataset for Chest X-ray Report Generation

Chest X-ray (CXR) reporting follows a region-based clinical workflow in which radiologists inspect anatomical regions and integrate localized findings into a final report. However, existing resources for CXR report generation provide these supervision signals in fragmented forms. We introduce MMRad-22K, a dataset that organizes regional textual observations, anatomical grounding coordinates, localized image evidence, and report targets into structured multimodal evidence units for CXR report generation. To motivate this formulation, we first compare different evidence formats for report generation and find that structured multimodal evidence is generally more useful than text-only or bounding box-based evidence. We then adapt a unified LVLM backbone using MMRad-22K and show that adaptation with multimodal evidence outperforms both textual-evidence adaptation and end-to-end adaptation on language and clinically oriented metrics. Under the same evaluation protocol, the adapted model also reaches a performance level comparable to several open-source LVLM references. Together, these results support MMRad-22K as a practical structured multimodal resource for training and evaluating CXR report generation aligned with clinical reading workflows.

cs.CV↗

arXiv2025

Super4DR: 4D Radar-centric Self-supervised Odometry and Gaussian-based Map Optimization

Conventional odometry and mapping methods using visual or LiDAR data often struggle under poor illumination and adverse weather conditions. Although 4D radar is suited for such environments, its sparse and noisy point clouds hinder accurate odometry estimation, while the radar maps suffer from obscure and incomplete structures. Thus, we propose Super4DR, a 4D radar-centric framework for learning-based odometry estimation and gaussian-based map optimization. First, we design a cluster-aware odometry network that incorporates object-level cues from the clustered radar points for inter-frame matching, alongside a hierarchical self-supervision mechanism to overcome outliers through spatio-temporal consistency, knowledge transfer, and feature contrast. Second, we propose using 3D gaussians as an intermediate representation, coupled with a radar-specific growth strategy, selective separation, and multi-view regularization, to recover blurry map areas and those undetected based on image texture. Experiments show that Super4DR achieves a 67% performance gain over prior self-supervised methods, nearly matches supervised odometry, and narrows the map quality disparity with LiDAR while enabling multi-modal image rendering.

cs.RO↗

arXiv2025

Asynchronous Nonlinear Sheaf Diffusion for Multi-Agent Coordination

Cellular sheaves and sheaf Laplacians provide a far-reaching generalization of graphs and graph Laplacians, resulting in a wide array of applications ranging from machine learning to multi-agent control. In the context of multi-agent systems, so called coordination sheaves provide a unifying formalism that models heterogeneous agents and coordination goals over undirected communication topologies, and applying sheaf diffusion drives agents to achieve their coordination goals. Existing literature on sheaf diffusion assumes that agents can communicate and compute updates synchronously, which is an unrealistic assumption in many scenarios where communication delays or heterogeneous agents with different compute capabilities cause disagreement among agents. To address these challenges, we introduce asynchronous nonlinear sheaf diffusion. Specifically, we show that under mild assumptions on the coordination sheaf and bounded delays in communication and computation, nonlinear sheaf diffusion converges to a minimizer of the Dirichlet energy of the coordination sheaf at a linear rate proportional to the delay bound. We further show that this linear convergence is attained from arbitrary initial conditions and the analysis depends on the spectrum of the sheaf Laplacian in a manner that generalizes the standard graph Laplacian case. We provide several numerical simulations to validate our theoretical results.

math.OC↗

arXiv2025

MARS2 2025 Challenge on Multimodal Reasoning: Datasets, Methods, Results, Discussion, and Outlook

This paper reviews the MARS2 2025 Challenge on Multimodal Reasoning. We aim to bring together different approaches in multimodal machine learning and LLMs via a large benchmark. We hope it better allows researchers to follow the state-of-the-art in this very dynamic area. Meanwhile, a growing number of testbeds have boosted the evolution of general-purpose large language models. Thus, this year's MARS2 focuses on real-world and specialized scenarios to broaden the multimodal reasoning applications of MLLMs. Our organizing team released two tailored datasets Lens and AdsQA as test sets, which support general reasoning in 12 daily scenarios and domain-specific reasoning in advertisement videos, respectively. We evaluated 40+ baselines that include both generalist MLLMs and task-specific models, and opened up three competition tracks, i.e., Visual Grounding in Real-world Scenarios (VG-RS), Visual Question Answering with Spatial Awareness (VQA-SA), and Visual Reasoning in Creative Advertisement Videos (VR-Ads). Finally, 76 teams from the renowned academic and industrial institutions have registered and 40+ valid submissions (out of 1200+) have been included in our ranking lists. Our datasets, code sets (40+ baselines and 15+ participants' methods), and rankings are publicly available on the MARS2 workshop website and our GitHub organization page https://github.com/mars2workshop/, where our updates and announcements of upcoming events will be continuously provided.

cs.CV↗

arXiv2025

A Distributed Asynchronous Generalized Momentum Algorithm Without Delay Bounds

Asynchronous optimization algorithms often require delay bounds to prove their convergence, though these bounds can be difficult to obtain in practice. Therefore, we introduce a novel distributed generalized momentum algorithm that provides fast convergence and allows arbitrary finite delays. It subsumes Nesterov's accelerated gradient algorithm and the heavy ball algorithm, among others. We first develop conditions on the parameters of the algorithm that ensure asymptotic convergence. Then we show its convergence rate is linear in a function of the number of computations and communications that processors perform (in a way that we make precise). Simulations compare this algorithm to gradient descent, heavy ball, and Nesterov's accelerated gradient algorithm with a text classification problem on the 20 newsgroups dataset. Across a range of scenarios with unbounded delays, the generalized momentum algorithm converges with at least 36% fewer iterations than gradient descent, 28% fewer iterations than the heavy ball algorithm, and 16% fewer iterations than Nesterov's accelerated gradient algorithm.

math.OC↗

arXiv2025

NEARL: Interacted Query Adaptation with Orthogonal Regularization for Medical Vision-Language Understanding

Computer-aided medical image analysis is crucial for disease diagnosis and treatment planning. While vision-language models (VLMs) such as CLIP exhibit strong generalization ability, their direct application to medical imaging remains hindered by a substantial domain gap. Existing methods for bridging this gap, including prompt learning and unidirectional modality interaction, typically introduce domain knowledge into only one modality. However, such approaches fail to fully exploit CLIP's inherent dual-modality structure and overlook the synergistic effect of bidirectional cross-modal interaction, resulting in persistent modality misalignment. In this paper, we propose NEARL (iNteracted quEry Adaptation with oRthogonaL Regularization), a novel parameter-efficient VLM framework for bidirectional cross-modal interaction. NEARL consists of two key components: (1) the Unified Synergy Embedding Transformer (USEformer), which dynamically generates compact cross-modal queries to facilitate interaction; and (2) the Orthogonal Cross-Attention Adapter (OCA), which decouples new knowledge into truly novel and incremental components through orthogonal regularization. This design reduces interference from incremental components, enabling more focused learning of novel information and improving modality interaction in VLMs. Notably, NEARL introduces only 1.46M learnable parameters. Extensive experiments on three medical imaging modalities demonstrate state-of-the-art performance (e.g., a 2.3% relative improvement on the pneumonia dataset), along with fast inference and low memory overhead, highlighting its effectiveness for real-world medical vision-language understanding.

cs.CV↗

arXiv2025

MedSeg-R: Reasoning Segmentation in Medical Images with Multimodal Large Language Models

Medical image segmentation is crucial for clinical diagnosis, yet existing models are limited by their reliance on explicit human instructions and lack the active reasoning capabilities to understand complex clinical questions. While recent advancements in multimodal large language models (MLLMs) have improved medical question-answering (QA) tasks, most methods struggle to generate precise segmentation masks, limiting their application in automatic medical diagnosis. In this paper, we introduce medical image reasoning segmentation, a novel task that aims to generate segmentation masks based on complex and implicit medical instructions. To address this, we propose MedSeg-R, an end-to-end framework that leverages the reasoning abilities of MLLMs to interpret clinical questions while also capable of producing corresponding precise segmentation masks for medical images. It is built on two core components: 1) a global context understanding module that interprets images and comprehends complex medical instructions to generate multi-modal intermediate tokens, and 2) a pixel-level grounding module that decodes these tokens to produce precise segmentation masks and textual responses. Furthermore, we introduce MedSeg-QA, a large-scale dataset tailored for the medical image reasoning segmentation task. It includes over 10,000 image-mask pairs and multi-turn conversations, automatically annotated using large language models and refined through physician reviews. Experiments show MedSeg-R's superior performance across several benchmarks, achieving high segmentation accuracy and enabling interpretable textual analysis of medical images.

cs.CV↗

arXiv2025

Integrating Natural Language Processing and Exercise Monitoring for Early Diagnosis of Metabolic Syndrome: A Deep Learning Approach

Metabolic syndrome (MetS) is a medication condition characterized by abdominal obesity, insulin resistance, hypertension and hyperlipidemia. It increases the risk of majority of chronic diseases, including type 2 diabetes mellitus, and affects about one quarter of the global population. Therefore, early detection and timely intervention for MetS are crucial. Standard diagnosis for MetS components requires blood tests conducted within medical institutions. However, it is frequently underestimated, leading to unmet need for care for MetS population. This study aims to use the least physiological data and free texts about exercises related activities, which are obtained easily in daily life, to diagnosis MetS. We collected the data from 40 volunteers in a nursing home and used data augmentation to reduce the imbalance. We propose a deep learning framework for classifying MetS that integrates natural language processing (NLP) and exercise monitoring. The results showed that the best model reported a high positive result (AUROC=0.806 and REC=76.3%) through 3-fold cross-validation. Feature importance analysis revealed that text and minimum heart rate on a daily basis contribute the most in the classification of MetS. This study demonstrates the potential application of data that are easily measurable in daily life for the early diagnosis of MetS, which could contribute to reducing the cost of screening and management for MetS population.

cs.AI↗

arXiv2024

Visual Grounding with Multi-modal Conditional Adaptation

Visual grounding is the task of locating objects specified by natural language expressions. Existing methods extend generic object detection frameworks to tackle this task. They typically extract visual and textual features separately using independent visual and textual encoders, then fuse these features in a multi-modal decoder for final prediction. However, visual grounding presents unique challenges. It often involves locating objects with different text descriptions within the same image. Existing methods struggle with this task because the independent visual encoder produces identical visual features for the same image, limiting detection performance. Some recently approaches propose various language-guided visual encoders to address this issue, but they mostly rely solely on textual information and require sophisticated designs. In this paper, we introduce Multi-modal Conditional Adaptation (MMCA), which enables the visual encoder to adaptively update weights, directing its focus towards text-relevant regions. Specifically, we first integrate information from different modalities to obtain multi-modal embeddings. Then we utilize a set of weighting coefficients, which generated from the multimodal embeddings, to reorganize the weight update matrices and apply them to the visual encoder of the visual grounding model. Extensive experiments on four widely used datasets demonstrate that MMCA achieves significant improvements and state-of-the-art results. Ablation experiments further demonstrate the lightweight and efficiency of our method. Our source code is available at: https://github.com/Mr-Bigworth/MMCA.

cs.CV↗

arXiv2024

An overview of domain-specific foundation model: key technologies, applications and challenges

The impressive performance of ChatGPT and other foundation-model-based products in human language understanding has prompted both academia and industry to explore how these models can be tailored for specific industries and application scenarios. This process, known as the customization of domain-specific foundation models (FMs), addresses the limitations of general-purpose models, which may not fully capture the unique patterns and requirements of domain-specific data. Despite its importance, there is a notable lack of comprehensive overview papers on building domain-specific FMs, while numerous resources exist for general-purpose models. To bridge this gap, this article provides a timely and thorough overview of the methodology for customizing domain-specific FMs. It introduces basic concepts, outlines the general architecture, and surveys key methods for constructing domain-specific models. Furthermore, the article discusses various domains that can benefit from these specialized models and highlights the challenges ahead. Through this overview, we aim to offer valuable guidance and reference for researchers and practitioners from diverse fields to develop their own customized FMs.

cs.AI↗

arXiv2024

Predicting Chaotic System Behavior using Machine Learning Techniques

Recently, machine learning techniques, particularly deep learning, have demonstrated superior performance over traditional time series forecasting methods across various applications, including both single-variable and multi-variable predictions. This study aims to investigate the capability of i) Next Generation Reservoir Computing (NG-RC) ii) Reservoir Computing (RC) iii) Long short-term Memory (LSTM) for predicting chaotic system behavior, and to compare their performance in terms of accuracy, efficiency, and robustness. These methods are applied to predict time series obtained from four representative chaotic systems including Lorenz, R\"ossler, Chen, Qi systems. In conclusion, we found that NG-RC is more computationally efficient and offers greater potential for predicting chaotic system behavior.

cs.LG↗

arXiv2023

Parallel in-memory wireless computing

Parallel wireless digital communication with ultralow power consumption is critical for emerging edge technologies such as 5G and Internet of Things. However, the physical separation between digital computing units and analogue transmission units in traditional wireless technology leads to high power consumption. Here we report a parallel in-memory wireless computing scheme. The approach combines in-memory computing with wireless communication using memristive crossbar arrays. We show that the system can be used for the radio transmission of a binary stream of 480 bits with a bit error rate of 0. The in-memory wireless computing uses two orders of magnitude less power than conventional technology (based on digital-to-analogue and analogue-to-digital converters). We also show that the approach can be applied to acoustic and optical wireless communications

cs.AR↗

arXiv2021

Scalable massively parallel computing using continuous-time data representation in nanoscale crossbar array

The growth of connected intelligent devices in the Internet of Things has created a pressing need for real-time processing and understanding of large volumes of analogue data. The difficulty in boosting the computing speed renders digital computing unable to meet the demand for processing analogue information that is intrinsically continuous in magnitude and time. By utilizing a continuous data representation in a nanoscale crossbar array, parallel computing can be implemented for the direct processing of analogue information in real time. Here, we propose a scalable massively parallel computing scheme by exploiting a continuous-time data representation and frequency multiplexing in a nanoscale crossbar array. This computing scheme enables the parallel reading of stored data and the one-shot operation of matrix-matrix multiplications in the crossbar array. Furthermore, we achieve the one-shot recognition of 16 letter images based on two physically interconnected crossbar arrays and demonstrate that the processing and modulation of analogue information can be simultaneously performed in a memristive crossbar array.

cond-mat.mtrl-sci↗