arXiv subjects

Jiawei Wang

Publications and source records attributed to Jiawei Wang.

At least 73 records · Page 4Linked to original sources

arXiv2026

InfiniDepth: Arbitrary-Resolution and Fine-Grained Depth Estimation with Neural Implicit Fields

Existing depth estimation methods are fundamentally limited to predicting depth on discrete image grids. Such representations restrict their scalability to arbitrary output resolutions and hinder the geometric detail recovery. This paper introduces InfiniDepth, which represents depth as neural implicit fields. Through a simple yet effective local implicit decoder, we can query depth at continuous 2D coordinates, enabling arbitrary-resolution and fine-grained depth estimation. To better assess our method's capabilities, we curate a high-quality 4K synthetic benchmark from five different games, spanning diverse scenes with rich geometric and appearance details. Extensive experiments demonstrate that InfiniDepth achieves state-of-the-art performance on both synthetic and real-world benchmarks across relative and metric depth estimation tasks, particularly excelling in fine-detail regions. It also benefits the task of novel view synthesis under large viewpoint shifts, producing high-quality results with fewer holes and artifacts.

cs.CV↗

arXiv2026

DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning

General reasoning represents a long-standing and formidable challenge in artificial intelligence. Recent breakthroughs, exemplified by large language models (LLMs) and chain-of-thought prompting, have achieved considerable success on foundational reasoning tasks. However, this success is heavily contingent upon extensive human-annotated demonstrations, and models' capabilities are still insufficient for more complex problems. Here we show that the reasoning abilities of LLMs can be incentivized through pure reinforcement learning (RL), obviating the need for human-labeled reasoning trajectories. The proposed RL framework facilitates the emergent development of advanced reasoning patterns, such as self-reflection, verification, and dynamic strategy adaptation. Consequently, the trained model achieves superior performance on verifiable tasks such as mathematics, coding competitions, and STEM fields, surpassing its counterparts trained via conventional supervised learning on human demonstrations. Moreover, the emergent reasoning patterns exhibited by these large-scale models can be systematically harnessed to guide and enhance the reasoning capabilities of smaller models.

cs.CL↗

arXiv2025

BALI: Branch-Aware Loop Invariant Inference with Large Language Models

Loop invariants are fundamental for reasoning about the correctness of iterative algorithms. However, deriving suitable invariants remains a challenging and often manual task, particularly for complex programs. In this paper, we introduce BALI, a branch-aware framework that integrates large language models (LLMs) to enhance the inference and verification of loop invariants. Our approach combines automated reasoning with branch-aware static program analysis to improve both precision and scalability. Specifically, unlike prior LLM-only guess-and-check methods, BALI first verifies branch-sequence-level (path-level) clauses with SMT and then composes them into program-level invariants. We outline its key components, present preliminary results, and discuss future directions toward fully automated invariant discovery.

cs.PL↗

arXiv2025

Towards Resilient Transportation: A Conditional Transformer for Accident-Informed Traffic Forecasting

Traffic prediction remains a key challenge in spatio-temporal data mining, despite progress in deep learning. Accurate forecasting is hindered by the complex influence of external factors such as traffic accidents and regulations, often overlooked by existing models due to limited data integration. To address these limitations, we present two enriched traffic datasets from Tokyo and California, incorporating traffic accident and regulation data. Leveraging these datasets, we propose ConFormer (Conditional Transformer), a novel framework that integrates graph propagation with guided normalization layer. This design dynamically adjusts spatial and temporal node relationships based on historical patterns, enhancing predictive accuracy. Our model surpasses the state-of-the-art STAEFormer in both predictive performance and efficiency, achieving lower computational costs and reduced parameter demands. Extensive evaluations demonstrate that ConFormer consistently outperforms mainstream spatio-temporal baselines across multiple metrics, underscoring its potential to advance traffic prediction research.

cs.LG↗

arXiv2025

Triplet pairing in neutron matter in a comprehensive diagrammatic approach

We apply a large-scale summation of Feynman diagrams, including the class of parquet diagrams plus important contributions outside the parquet class, for calculating effective pairing interactions and subsequently the superfluid gap in P-wave pairing in neutron matter. We use realistic nucleon-nucleon interactions of the $v_8$ type and perform calculations up to a Fermi momentum of 1.8 fm$^{-1}$. We find that many-body correlations lead to a strong reduction of the spin-orbit interaction, and, therefore, to a radical suppression of the $^3P_2-^3F_2$ gap and an enhancement of the $^3P_0$ gap.

nucl-th↗

arXiv2025

Mixed Platoon Control under Noise and Attacks: Robust Data-Driven Predictive Control and Human-in-the-Loop Validation

Controlling mixed platoons, which consist of both connected and automated vehicles (CAVs) and human-driven vehicles (HDVs), poses significant challenges due to the uncertain and unknown human driving behaviors. Data-driven control methods offer promising solutions by leveraging available trajectory data, but their performance can be compromised by noise and attacks. To address this issue, this paper proposes a Robust Data-EnablEd Predictive Leading Cruise Control (RDeeP-LCC) framework based on data-driven reachability analysis. The framework over-approximates system dynamics under noise and attack using a matrix zonotope set derived from data, and develops a stabilizing feedback control law. By decoupling the mixed platoon system into nominal and error components, we employ data-driven reachability sets to recursively compute error reachable sets that account for noise and attacks, and obtain tightened safety constraints of the nominal system. This leads to a robust data-driven predictive control framework, solved in a tube-based control manner. Human-in-the-loop experiments demonstrate that the RDeeP-LCC method significantly improves robustness against noise and attacks, while enhancing tracking accuracy, control efficiency, energy economy, driving comfort, and driving safety.

eess.SY↗

arXiv2025

Robust Nonlinear Data-Driven Predictive Control for Mixed Vehicle Platoons via Koopman Operator and Reachability Analysis

Mixed vehicle platoons, comprising connected and automated vehicles (CAVs) and human-driven vehicles (HDVs), hold significant potential for enhancing traffic performance. However, most existing control strategies assume linear system dynamics and often ignore the impact of adverse factors such as noise, disturbances, and attacks, which are inherent to real-world scenarios. To address these limitations, we propose a Robust Nonlinear Data-Driven Predictive Control (RNDDPC) framework that ensures safe and optimal control under uncertain and adverse conditions. By utilizing Koopman operator theory, we map the system's nonlinear dynamics into a higher-dimensional space, constructing a Koopman-based model that approximates the behavior of the original nonlinear system. To mitigate modeling errors associated with this predictor, we introduce a data-driven reachable set analysis technique that performs secondary learning using matrix zonotope sets, generating a reachable set predictor for over-approximation of the future states of the underlying system. Then, we formulate the RNDDPC optimization problem and solve it in a receding horizon manner for robust control inputs. Extensive simulations demonstrate that the proposed framework significantly outperforms baseline methods in tracking performance under noise, disturbances, and attacks.

eess.SY↗

arXiv2025

Tunable absorption spectrum splitting in a pulse-driven three-level system

When a two-level system is driven on resonance by a strong incident field, its emission spectrum is characterized by the well-known Mollow triplet. If the absorption from the excited state, in this continuously driven two-level system, to a third, higher energy level, is probed by a weak field, the resulting absorption spectrum features the Autler-Townes doublet with two peaks separated by the Rabi frequency of the strong driving field. It has been shown that when the two-level system is instead driven by a periodic pulse sequence, the emission spectrum obtained has similarities with the Mollow triplet even though the system is only driven during the short application time of the pulses and is allowed to evolve freely between pulses. Here, we evaluate the absorption spectrum of the three-level system in the ladder/cascade configuration when the bottom two levels are driven by a periodic pulse sequence while the transition between the middle and the highest level is probed by a weak field. The absorption spectrum displays similarities with the Autler-Townes doublet with frequency separation between the main peaks defined by the inter-pulse delay. In addition, this spectrum shows little dependence on the pulse carrier frequency. These results demonstrate the capacity to modulate the absorption spectrum of a three-level system with experimentally achievable pulse protocols.

physics.optics↗

arXiv2025

MCPMark: A Benchmark for Stress-Testing Realistic and Comprehensive MCP Use

MCP standardizes how LLMs interact with external systems, forming the foundation for general agents. However, existing MCP benchmarks remain narrow in scope: they focus on read-heavy tasks or tasks with limited interaction depth, and fail to capture the complexity and realism of real-world workflows. To address this gap, we propose MCPMark, a benchmark designed to evaluate MCP use in a more realistic and comprehensive manner. It consists of $127$ high-quality tasks collaboratively created by domain experts and AI agents. Each task begins with a curated initial state and includes a programmatic script for automatic verification. These tasks demand richer and more diverse interactions with the environment, involving a broad range of create, read, update, and delete (CRUD) operations. We conduct a comprehensive evaluation of cutting-edge LLMs using a minimal agent framework that operates in a tool-calling loop. Empirical results show that the best-performing model, gpt-5-medium, reaches only $52.56$\% pass@1 and $33.86$\% pass^4, while other widely regarded strong models, including claude-sonnet-4 and o3, fall below $30$\% pass@1 and $15$\% pass^4. On average, LLMs require $16.2$ execution turns and $17.4$ tool calls per task, significantly surpassing those in previous MCP benchmarks and highlighting the stress-testing nature of MCPMark.

cs.CL↗

arXiv2025

Hole spin splitting in a Ge quantum dot with finite barriers

We study the low-energy spectrum of a single hole confined in a planar Ge quantum dot (QD) within the effective-mass formalism. The QD is sandwiched between two GeSi barriers of finite potential height grown along the [001] direction. To treat this finite barrier problem, we adopt an independent-band approach in dealing with boundary conditions. The effects of different system parameters are investigated, including the width of the out-of-plane confining well, the size of the dot, and silicon concentration in the confining layers. The more accurate finite-barrier model results in the non-negligible dependence of the anisotropic $g$-factor on the choice of boundary conditions and on the silicon concentration in the barrier. Furthermore, while the ideal model of a planar dot with a square-well heterostructure already has an intrinsic spin-orbit coupling, realistic effects arising from the experimental setup may give rise to additional contributions. We investigate the impact from the top-gate electric field and the residual tensile strain on the qubit states. The results indicate that these effects are important contributions to the total spin-orbit coupling which enables fast electric control.

cond-mat.mes-hall↗

arXiv2025

TeraSim-World: Worldwide Safety-Critical Data Synthesis for End-to-End Autonomous Driving

Safe and scalable deployment of end-to-end (E2E) autonomous driving requires extensive and diverse data, particularly safety-critical events. Existing data are mostly generated from simulators with a significant sim-to-real gap or collected from on-road testing that is costly and unsafe. This paper presents TeraSim-World, an automated pipeline that synthesizes realistic and geographically diverse safety-critical data for E2E autonomous driving at anywhere in the world. Starting from an arbitrary location, TeraSim-World retrieves real-world maps and traffic demand from geospatial data sources. Then, it simulates agent behaviors from naturalistic driving datasets, and orchestrates diverse adversities to create corner cases. Informed by street views of the same location, it achieves photorealistic, geographically grounded sensor rendering via the frontier video generation model Cosmos-Drive. By bridging agent and sensor simulations, TeraSim-World provides a scalable and critical data synthesis framework for training and evaluation of E2E autonomous driving systems. Codes and videos are available at https://wjiawei.com/terasim-world-web/ .

cs.RO↗

arXiv2025

Harnessing Uncertainty: Entropy-Modulated Policy Gradients for Long-Horizon LLM Agents

In long-horizon tasks, recent agents based on Large Language Models (LLMs) face a significant challenge that sparse, outcome-based rewards make it difficult to assign credit to intermediate steps. Previous methods mainly focus on creating dense reward signals to guide learning, either through traditional reinforcement learning techniques like inverse reinforcement learning or by using Process Reward Models for step-by-step feedback. In this paper, we identify a fundamental problem in the learning dynamics of LLMs: the magnitude of policy gradients is inherently coupled with the entropy, which leads to inefficient small updates for confident correct actions and potentially destabilizes large updates for uncertain ones. To resolve this, we propose Entropy-Modulated Policy Gradients (EMPG), a framework that re-calibrates the learning signal based on step-wise uncertainty and the final task outcome. EMPG amplifies updates for confident correct actions, penalizes confident errors, and attenuates updates from uncertain steps to stabilize exploration. We further introduce a bonus term for future clarity that encourages agents to find more predictable solution paths. Through comprehensive experiments on three challenging agent tasks, WebShop, ALFWorld, and Deep Search, we demonstrate that EMPG achieves substantial performance gains and significantly outperforms strong policy gradient baselines. Project page is at https://empgseed-seed.github.io/

cs.LG↗

arXiv2025

WideSearch: Benchmarking Agentic Broad Info-Seeking

From professional research to everyday planning, many tasks are bottlenecked by wide-scale information seeking, which is more repetitive than cognitively complex. With the rapid development of Large Language Models (LLMs), automated search agents powered by LLMs offer a promising solution to liberate humans from this tedious work. However, the capability of these agents to perform such "wide-context" collection reliably and completely remains largely unevaluated due to a lack of suitable benchmarks. To bridge this gap, we introduce WideSearch, a new benchmark engineered to evaluate agent reliability on these large-scale collection tasks. The benchmark features 200 manually curated questions (100 in English, 100 in Chinese) from over 15 diverse domains, grounded in real user queries. Each task requires agents to collect large-scale atomic information, which could be verified one by one objectively, and arrange it into a well-organized output. A rigorous five-stage quality control pipeline ensures the difficulty, completeness, and verifiability of the dataset. We benchmark over 10 state-of-the-art agentic search systems, including single-agent, multi-agent frameworks, and end-to-end commercial systems. Most systems achieve overall success rates near 0\%, with the best performer reaching just 5\%. However, given sufficient time, cross-validation by multiple human testers can achieve a near 100\% success rate. These results demonstrate that present search agents have critical deficiencies in large-scale information seeking, underscoring urgent areas for future research and development in agentic search. Our dataset, evaluation pipeline, and benchmark results have been publicly released at https://widesearch-seed.github.io/

cs.CL↗

arXiv2025

Human-Machine Shared Control Approach for the Takeover of CACC

Cooperative Adaptive Cruise Control (CACC) often requires human takeover for tasks such as exiting a freeway. Direct human takeover can pose significant risks, especially given the close-following strategy employed by CACC, which might cause drivers to feel unsafe and execute hard braking, potentially leading to collisions. This research aims to develop a CACC takeover controller that ensures a smooth transition from automated to human control. The proposed CACC takeover maneuver employs an indirect human-machine shared control approach, modeled as a Stackelberg competition where the machine acts as the leader and the human as the follower. The machine guides the human to respond in a manner that aligns with the machine's expectations, aiding in maintaining following stability. Additionally, the human reaction function is integrated into the machine's predictive control system, moving beyond a simple "prediction-planning" pipeline to enhance planning optimality. The controller has been verified to i) enable a smooth takeover maneuver of CACC; ii) ensure string stability in the condition that the platoon has less than 6 CAVs and human control authority is less than 40%; iii) enhance both perceived and actual safety through machine interventions; and iv) reduce the impact on upstream traffic by up to 60%.

cs.RO↗

arXiv2025

Length Matters: Length-Aware Transformer for Temporal Sentence Grounding

Temporal sentence grounding (TSG) is a highly challenging task aiming to localize the temporal segment within an untrimmed video corresponding to a given natural language description. Benefiting from the design of learnable queries, the DETR-based models have achieved substantial advancements in the TSG task. However, the absence of explicit supervision often causes the learned queries to overlap in roles, leading to redundant predictions. Therefore, we propose to improve TSG by making each query fulfill its designated role, leveraging the length priors of the video-description pairs. In this paper, we introduce the Length-Aware Transformer (LATR) for TSG, which assigns different queries to handle predictions based on varying temporal lengths. Specifically, we divide all queries into three groups, responsible for segments with short, middle, and long temporal durations, respectively. During training, an additional length classification task is introduced. Predictions from queries with mismatched lengths are suppressed, guiding each query to specialize in its designated function. Extensive experiments demonstrate the effectiveness of our LATR, achieving state-of-the-art performance on three public benchmarks. Furthermore, the ablation studies validate the contribution of each component of our method and the critical role of incorporating length priors into the TSG task.

cs.CV↗

arXiv2025

Safeguarding Vision-Language Models: Mitigating Vulnerabilities to Gaussian Noise in Perturbation-based Attacks

Vision-Language Models (VLMs) extend the capabilities of Large Language Models (LLMs) by incorporating visual information, yet they remain vulnerable to jailbreak attacks, especially when processing noisy or corrupted images. Although existing VLMs adopt security measures during training to mitigate such attacks, vulnerabilities associated with noise-augmented visual inputs are overlooked. In this work, we identify that missing noise-augmented training causes critical security gaps: many VLMs are susceptible to even simple perturbations such as Gaussian noise. To address this challenge, we propose Robust-VLGuard, a multimodal safety dataset with aligned / misaligned image-text pairs, combined with noise-augmented fine-tuning that reduces attack success rates while preserving functionality of VLM. For stronger optimization-based visual perturbation attacks, we propose DiffPure-VLM, leveraging diffusion models to convert adversarial perturbations into Gaussian-like noise, which can be defended by VLMs with noise-augmented safety fine-tuning. Experimental results demonstrate that the distribution-shifting property of diffusion model aligns well with our fine-tuned VLMs, significantly mitigating adversarial perturbations across varying intensities. The dataset and code are available at https://github.com/JarvisUSTC/DiffPure-RobustVLM.

cs.CV↗

arXiv2025

On Improving PPG-Based Sleep Staging: A Pilot Study

Sleep monitoring through accessible wearable technology is crucial to improving well-being in ubiquitous computing. Although photoplethysmography(PPG) sensors are widely adopted in consumer devices, achieving consistently reliable sleep staging using PPG alone remains a non-trivial challenge. In this work, we explore multiple strategies to enhance the performance of PPG-based sleep staging. Specifically, we compare conventional single-stream model with dual-stream cross-attention strategies, based on which complementary information can be learned via PPG and PPG-derived modalities such as augmented PPG or synthetic ECG. To study the effectiveness of the aforementioned approaches in four-stage sleep monitoring task, we conducted experiments on the world's largest sleep staging dataset, i.e., the Multi-Ethnic Study of Atherosclerosis(MESA). We found that substantial performance gain can be achieved by combining PPG and its auxiliary information under the dual-stream cross-attention architecture. Source code of this project can be found at https://github.com/DavyWJW/sleep-staging-models

eess.SP↗

arXiv2025

Photonic chip-based optical frequency division with PZT-integrated soliton microcombs

Optical frequency division (OFD) produces low-noise microwave and millimeter-wave signals by transferring the exceptional stability of optical references to electronic frequency domains. Recent developments in integrated optical references and soliton microcombs have paved the way for miniaturizing OFD oscillators to chip scale. Critical to this realization is a rapid tunable frequency comb that is stabilized to the optical references, thereby coherently linking optical and electronic frequencies. In this work, we advance the on-chip OFD technology using an integrated high-speed PZT stress-optic actuator on the SiN soliton microcomb resonator. The integrated PZT actuator tunes the resonance frequency of the soliton-generating microresonator with a bandwidth exceeding 10s MHz and independently adjusts the soliton repetition rate without perturbing the frequency comb offset. Optical frequency division and low-noise mmWave generation are demonstrated by feedback control of the soliton repetition rate through the integrated PZT-actuator, and the soliton microcomb is stabilized to a pair of reference lasers that are locked to an integrated 4-meter SiN coil reference cavity. Our approach provides a fast, versatile and integrated control mechanism for OFD oscillators and their applications in advanced communications, sensing, and precise timing.

physics.optics↗