arXiv subjects

Zhiyuan Zhou

Publications and source records attributed to Zhiyuan Zhou.

At least 19 recordsLinked to original sources

arXiv2026

Fluctuation-Driven Nonlinear Amplification of Quantum Statistics

Photon statistics have moved to the forefront of modern optics, as intensity fluctuations and correlations shape multiphoton interactions and reveal information beyond mean-intensity measurements. Developing high-quality photon sources with pronounced correlations is a fundamental necessity in these fields. Here we demonstrate fluctuation-driven nonlinear statistical amplification of quantum light in spontaneous four-wave mixing using filtered amplified spontaneous emission (ASE). Extending the coherent-pump framework to fluctuating fields, we show how nonlinear weighting of pump intensity combines with bosonic bunching to amplify quantum statistics and reshape temporal correlations. In a SiN microring, ASE pumping increases the zero-delay unconditional second-order correlation from 2.01 to 7.58 and extends the Hanbury Brown--Twiss correlation time by a factor of approximately 2.4. The super-bunched quantum source nevertheless retains heralded single-photon behaviour with $g_H^{(2)}(0)\simeq0.04$, while the same ASE pump supports time--energy entanglement in a silicon waveguide with a raw Franson visibility of 89.84\%. These results establish driving-field statistics as a design dimension for quantum light, broadening the horizons for research into higher-order correlations and nonlinear physics.

quant-ph↗

arXiv2026

Any to Full: Prompting Depth Anything for Depth Completion in One Stage

Accurate, dense depth estimation is crucial for robotic perception, but commodity sensors often yield sparse or incomplete measurements due to hardware limitations. Existing RGBD-fused depth completion methods learn priors jointly conditioned on training RGB distribution and specific depth patterns, limiting domain generalization and robustness to various depth patterns. Recent efforts leverage monocular depth estimation (MDE) models to introduce domain-general geometric priors, but current two-stage integration strategies relying on explicit relative-to-metric alignment incur additional computation and introduce structured distortions. To this end, we present Any2Full, a one-stage, domain-general, and pattern-agnostic framework that reformulates completion as a scale-prompting adaptation of a pretrained MDE model. To address varying depth sparsity levels and irregular spatial distributions, we design a Scale-Aware Prompt Encoder. It distills scale cues from sparse inputs into unified scale prompts, guiding the MDE model toward globally scale-consistent predictions while preserving its geometric priors. Extensive experiments demonstrate that Any2Full achieves superior robustness and efficiency. It outperforms OMNI-DC by 32.2\% in average AbsREL and delivers a 1.4$\times$ speedup over PriorDA with the same MDE backbone, establishing a new paradigm for universal depth completion. Codes and checkpoints are available at https://github.com/zhiyuandaily/Any2Full.

cs.CV↗

arXiv2026

A knowledge-augmented dataset of high-risk driving scenarios with LLM annotations for autonomous driving

Safe autonomous driving requires both rapid responses to common high-risk events and deeper reasoning over rare, extreme long-tail scenarios in traffic safety. These scenarios are severely under-represented in naturalistic driving data, and existing trajectory and language-augmented datasets seldom provide high-risk event labels, semantic annotations, and verifiable safety signals. Here we present K-Risk, a knowledge-augmented dataset that combines structured driving trajectories with large language model generated semantic annotations for safety-critical driving scenarios. K-Risk integrates 20 human-driven and autonomous-vehicle trajectory datasets from Europe, China, and the United States, covering highways, urban freeways, intersections, and roundabouts. Using a unified risk-centric extraction pipeline, K-Risk curates 31,398 high-risk events, together with a 1,036-event extreme subset of near-collision cases. Each event is released as a synchronized trajectory, metadata, and language triplet containing structured scenario descriptions, abnormal-behavior notifications, and, for a representative subset, causal risk analyses and action recommendations validated through a closed-loop simulator with iterative reflection. By combining multi-dimensional risk annotations, interpretable language supervision, and verifiable decisions, K-Risk bridges structured traffic trajectories, semantic reasoning, and decision supervision, providing a standardized foundation for developing and evaluating next-generation risk-aware autonomous driving agents.

cs.LG↗

arXiv2026

Test-Time Gradient Guidance of Flow Policies in Reinforcement Learning

Expressive continuous control policies, such as diffusion and flow models, form the backbone of recent advances in scaling imitation learning for simulated and real robot control. While they are known to scale stably in the supervised imitation learning setting, incorporating them into reinforcement learning (RL) pipelines for policy improvement has proven more difficult. It often requires specialized training objectives or backpropagating through denoising processes, which cause well-known issues with stability and affect scalability. In this paper we study the question of whether simple policy improvement schemes at test time alone, leaving stable supervised policy training intact, can be a competitive alternative which sidesteps these issues. To this end, we propose QGF (Q-Guided Flow), an RL algorithm that performs policy optimization entirely at test time. QGF works by pre-training both a reference flow policy (via a standard behavioral cloning objective) and a value function critic and, at test time, using the value gradient to guide the reference policy to generate higher-value actions without any additional policy learning. Empirically, QGF outperforms prior test-time RL methods on single-task and goal-conditioned offline RL benchmarks with high-dimensional action spaces, and is competitive with state-of-the-art training-time algorithms while being much cheaper to run. Moreover, it exhibits favorable scaling with model size by avoiding the instability of actor-critic training, offering a practical and effective alternative RL algorithm with expressive policies.

cs.LG↗

arXiv2026

Reinforcement Learning with Action Chunking

We present Q-chunking, a simple yet effective recipe for improving reinforcement learning (RL) algorithms for long-horizon, sparse-reward tasks. Our recipe is designed for the offline-to-online RL setting, where the goal is to leverage an offline prior dataset to maximize the sample-efficiency of online learning. Effective exploration and sample-efficient learning remain central challenges in this setting, as it is not obvious how the offline data should be utilized to acquire a good exploratory policy. Our key insight is that action chunking, a technique popularized in imitation learning where sequences of future actions are predicted rather than a single action at each timestep, can be applied to temporal difference (TD)-based RL methods to mitigate the exploration challenge. Q-chunking adopts action chunking by directly running RL in a 'chunked' action space, enabling the agent to (1) leverage temporally consistent behaviors from offline data for more effective online exploration and (2) use unbiased $n$-step backups for more stable and efficient TD learning. Our experimental results demonstrate that Q-chunking exhibits strong offline performance and online sample efficiency, outperforming prior best offline-to-online methods on a range of long-horizon, sparse-reward manipulation tasks.

cs.LG↗

arXiv2026

CausalVAD: De-confounding End-to-End Autonomous Driving via Causal Intervention

Planning-oriented end-to-end driving models show great promise, yet they fundamentally learn statistical correlations instead of true causal relationships. This vulnerability leads to causal confusion, where models exploit dataset biases as shortcuts, critically harming their reliability and safety in complex scenarios. To address this, we introduce CausalVAD, a de-confounding training framework that leverages causal intervention. At its core, we design the sparse causal intervention scheme (SCIS), a lightweight, plug-and-play module to instantiate the backdoor adjustment theory in neural networks. SCIS constructs a dictionary of prototypes representing latent driving contexts. It then uses this dictionary to intervene on the model's sparse vectorized queries. This step actively eliminates spurious associations induced by confounders, thereby eliminating spurious factors from the representations for downstream tasks. Extensive experiments on benchmarks like nuScenes show CausalVAD achieves state-of-the-art planning accuracy and safety. Furthermore, our method demonstrates superior robustness against both data bias and noisy scenarios configured to induce causal confusion.

cs.CV↗

arXiv2026

Differential Mental Disorder Detection with Psychology-Inspired Multimodal Stimuli

Differential diagnosis of mental disorders remains a fundamental challenge in real-world clinical practice, where multiple conditions often exhibit overlapping symptoms. However, most existing public datasets are developed under single-disorder settings and rely on limited data elicitation paradigms, restricting their ability to capture disorder-specific patterns. In this work, we investigate differential mental disorder detection through psychology-inspired multimodal stimuli, designed to elicit diverse emotional, cognitive, and behavioral responses grounded in findings from experimental psychology. Based on this paradigm, we collect a large-scale multimodal mental health dataset (MMH) covering depression, anxiety, and schizophrenia, with all diagnostic labels clinically verified by licensed psychiatrists. To effectively model the heterogeneous signals induced by diverse elicitation tasks, we further propose a paradigm-aware multimodal framework that leverages inter-disorder differences prior knowledge as prompt-guided semantic descriptions to capture task-specific affective and interaction contexts for multimodal representation learning in the new differential mental disorder detection task. Extensive experiments show that our framework consistently outperforms existing baselines, underscoring the value of psychology-inspired stimulus design for differential mental disorder detection.

cs.MM↗

arXiv2026

InterMind: Doctor-Patient-Family Interactive Depression Assessment Empowered by Large Language Models

Depression poses significant challenges to patients and healthcare organizations, necessitating efficient assessment methods. Existing paradigms typically focus on a patient-doctor way that overlooks multi-role interactions, such as family involvement in the evaluation and caregiving process. Moreover, current automatic depression detection (ADD) methods usually model depression detection as a classification or regression task, lacking interpretability for the decision-making process. To address these issues, we developed InterMind, a doctor-patient-family interactive depression assessment system empowered by large language models (LLMs). Our system enables patients and families to contribute descriptions, generates assistive diagnostic reports for doctors, and provides actionable insights, improving diagnostic precision and efficiency. To enhance LLMs' performance in psychological counseling and diagnostic interpretability, we integrate retrieval-augmented generation (RAG) and chain-of-thoughts (CoT) techniques for data augmentation, which mitigates the hallucination issue of LLMs in specific scenarios after instruction fine-tuning. Quantitative experiments and professional assessments by clinicians validate the effectiveness of our system.

cs.HC↗

arXiv2026

Voices, Faces, and Feelings: Multi-modal Emotion-Cognition Captioning for Mental Health Understanding

Emotional and cognitive factors are essential for understanding mental health disorders. However, existing methods often treat multi-modal data as classification tasks, limiting interpretability especially for emotion and cognition. Although large language models (LLMs) offer opportunities for mental health analysis, they mainly rely on textual semantics and overlook fine-grained emotional and cognitive cues in multi-modal inputs. While some studies incorporate emotional features via transfer learning, their connection to mental health conditions remains implicit. To address these issues, we propose ECMC, a novel task that aims at generating natural language descriptions of emotional and cognitive states from multi-modal data, and producing emotion-cognition profiles that improve both the accuracy and interpretability of mental health assessments. We adopt an encoder-decoder architecture, where modality-specific encoders extract features, which are fused by a dual-stream BridgeNet based on Q-former. Contrastive learning enhances the extraction of emotional and cognitive features. A LLaMA decoder then aligns these features with annotated captions to produce detailed descriptions. Extensive objective and subjective evaluations demonstrate that: 1) ECMC outperforms existing multi-modal LLMs and mental health models in generating emotion-cognition captions; 2) the generated emotion-cognition profiles significantly improve assistive diagnosis and interpretability in mental health analysis.

cs.MM↗

arXiv2026

Robust Finetuning of Vision-Language-Action Robot Policies via Parameter Merging

Generalist robot policies, trained on large and diverse datasets, have demonstrated the ability to generalize across a wide spectrum of behaviors, enabling a single policy to act in varied real-world environments. However, they still fall short on new tasks not covered in the training data. When finetuned on limited demonstrations of a new task, these policies often overfit to the specific demonstrations--not only losing their prior abilities to solve a wide variety of generalist tasks but also failing to generalize within the new task itself. In this work, we aim to develop a method that preserves the generalization capabilities of the generalist policy during finetuning, allowing a single policy to robustly incorporate a new skill into its repertoire. Our goal is a single policy that both learns to generalize to variations of the new task and retains the broad competencies gained from pretraining. We show that this can be achieved through a simple yet effective strategy: interpolating the weights of a finetuned model with that of the pretrained model. We show, across extensive simulated and real-world experiments, that such model merging produces a single model that inherits the generalist abilities of the base model and learns to solve the new task robustly, outperforming both the pretrained and finetuned model on out-of-distribution variations of the new task. Moreover, we show that model merging performance scales with the amount of pretraining data, and enables continual acquisition of new skills in a lifelong learning setting, without sacrificing previously learned generalist abilities.

cs.RO↗

arXiv2025

Learning Spatial-Aware Manipulation Ordering

Manipulation in cluttered environments is challenging due to spatial dependencies among objects, where an improper manipulation order can cause collisions or blocked access. Existing approaches often overlook these spatial relationships, limiting their flexibility and scalability. To address these limitations, we propose OrderMind, a unified spatial-aware manipulation ordering framework that directly learns object manipulation priorities based on spatial context. Our architecture integrates a spatial context encoder with a temporal priority structuring module. We construct a spatial graph using k-Nearest Neighbors to aggregate geometric information from the local layout and encode both object-object and object-manipulator interactions to support accurate manipulation ordering in real-time. To generate physically and semantically plausible supervision signals, we introduce a spatial prior labeling method that guides a vision-language model to produce reasonable manipulation orders for distillation. We evaluate OrderMind on our Manipulation Ordering Benchmark, comprising 163,222 samples of varying difficulty. Extensive experiments in both simulation and real-world environments demonstrate that our method significantly outperforms prior approaches in effectiveness and efficiency, enabling robust manipulation in cluttered scenes.

cs.RO↗

arXiv2025

$π^{*}_{0.6}$: a VLA That Learns From Experience

We study how vision-language-action (VLA) models can improve through real-world deployments via reinforcement learning (RL). We present a general-purpose method, RL with Experience and Corrections via Advantage-conditioned Policies (RECAP), that provides for RL training of VLAs via advantage conditioning. Our method incorporates heterogeneous data into the self-improvement process, including demonstrations, data from on-policy collection, and expert teleoperated interventions provided during autonomous execution. RECAP starts by pre-training a generalist VLA with offline RL, which we call $π^{*}_{0.6}$, that can then be specialized to attain high performance on downstream tasks through on-robot data collection. We show that the $π^{*}_{0.6}$ model trained with the full RECAP method can fold laundry in real homes, reliably assemble boxes, and make espresso drinks using a professional espresso machine. On some of the hardest tasks, RECAP more than doubles task throughput and roughly halves the task failure rate.

cs.LG↗

arXiv2025

Short-wave infrared broadband up-conversion imaging by using a noncritical phase matched bulk KTiOPO$_4$ crystal

Compared to cryogenically cooled conventional detectors, up-conversion detection enables efficient room-temperature short-wave infrared (SWIR) imaging. Although quasi-phase-matching (QPM) in periodically poled crystals offers advantages, the small crystal aperture (typically 1 mm$\times$3 mm) limits resolution. Non-poled crystals enable larger apertures but suffer walk-off aberrations. This work overcomes these limitations by using a noncritical phase matched (NCPM) KTiOPO$_4$ crystal (6 mm$\times$7 mm aperture, 0.5 mm length). Results show resolutions 6$\times$ and 2$\times$ higher than periodically poled crystals in orthogonal directions, with broad conversion band (1.3-2.2 $μ$m) covering biological and atmospheric windows. The absence of walk-off ensures better image fidelity in up-conversion process. This study presents the first comprehensive characterization of NCPM-based broadband up-conversion imaging, demonstrating performance at the theoretical resolution limit while circumventing drawbacks inherent in alternative up-conversion schemes and conventional detectors.

physics.optics↗

arXiv2025

Compute-Optimal Scaling for Value-Based Deep RL

As models grow larger and training them becomes expensive, it becomes increasingly important to scale training recipes not just to larger models and more data, but to do so in a compute-optimal manner that extracts maximal performance per unit of compute. While such scaling has been well studied for language modeling, reinforcement learning (RL) has received less attention in this regard. In this paper, we investigate compute scaling for online, value-based deep RL. These methods present two primary axes for compute allocation: model capacity and the update-to-data (UTD) ratio. Given a fixed compute budget, we ask: how should resources be partitioned across these axes to maximize sample efficiency? Our analysis reveals a nuanced interplay between model size, batch size, and UTD. In particular, we identify a phenomenon we call TD-overfitting: increasing the batch quickly harms Q-function accuracy for small models, but this effect is absent in large models, enabling effective use of large batch size at scale. We provide a mental model for understanding this phenomenon and build guidelines for choosing batch size and UTD to optimize compute usage. Our findings provide a grounded starting point for compute-optimal scaling in deep RL, mirroring studies in supervised learning but adapted to TD learning.

cs.LG↗

arXiv2025

Behavioral Exploration: Learning to Explore via In-Context Adaptation

Developing autonomous agents that quickly explore an environment and adapt their behavior online is a canonical challenge in robotics and machine learning. While humans are able to achieve such fast online exploration and adaptation, often acquiring new information and skills in only a handful of interactions, existing algorithmic approaches tend to rely on random exploration and slow, gradient-based behavior updates. How can we endow autonomous agents with such capabilities on par with humans? Taking inspiration from recent progress on both in-context learning and large-scale behavioral cloning, in this work we propose behavioral exploration: training agents to internalize what it means to explore and adapt in-context over the space of ``expert'' behaviors. To achieve this, given access to a dataset of expert demonstrations, we train a long-context generative model to predict expert actions conditioned on a context of past observations and a measure of how ``exploratory'' the expert's behaviors are relative to this context. This enables the model to not only mimic the behavior of an expert, but also, by feeding its past history of interactions into its context, to select different expert behaviors than what have been previously selected, thereby allowing for fast online adaptation and targeted, ``expert-like'' exploration. We demonstrate the effectiveness of our method in both simulated locomotion and manipulation settings, as well as on real-world robotic manipulation tasks, illustrating its ability to learn adaptive, exploratory behavior.

cs.LG↗

arXiv2025

Efficient Online Reinforcement Learning Fine-Tuning Need Not Retain Offline Data

The modern paradigm in machine learning involves pre-training on diverse data, followed by task-specific fine-tuning. In reinforcement learning (RL), this translates to learning via offline RL on a diverse historical dataset, followed by rapid online RL fine-tuning using interaction data. Most RL fine-tuning methods require continued training on offline data for stability and performance. However, this is undesirable because training on diverse offline data is slow and expensive for large datasets, and in principle, also limit the performance improvement possible because of constraints or pessimism on offline data. In this paper, we show that retaining offline data is unnecessary as long as we use a properly-designed online RL approach for fine-tuning offline RL initializations. To build this approach, we start by analyzing the role of retaining offline data in online fine-tuning. We find that continued training on offline data is mostly useful for preventing a sudden divergence in the value function at the onset of fine-tuning, caused by a distribution mismatch between the offline data and online rollouts. This divergence typically results in unlearning and forgetting the benefits of offline pre-training. Our approach, Warm-start RL (WSRL), mitigates the catastrophic forgetting of pre-trained initializations using a very simple idea. WSRL employs a warmup phase that seeds the online RL run with a very small number of rollouts from the pre-trained policy to do fast online RL. The data collected during warmup helps ``recalibrate'' the offline Q-function to the online distribution, allowing us to completely discard offline data without destabilizing the online RL fine-tuning. We show that WSRL is able to fine-tune without retaining any offline data, and is able to learn faster and attains higher performance than existing algorithms irrespective of whether they retain offline data or not.

cs.LG↗

arXiv2025

REACT: Runtime-Enabled Active Collision-avoidance Technique for Autonomous Driving

Achieving rapid and effective active collision avoidance in dynamic interactive traffic remains a core challenge for autonomous driving. This paper proposes REACT (Runtime-Enabled Active Collision-avoidance Technique), a closed-loop framework that integrates risk assessment with active avoidance control. By leveraging energy transfer principles and human-vehicle-road interaction modeling, REACT dynamically quantifies runtime risk and constructs a continuous spatial risk field. The system incorporates physically grounded safety constraints such as directional risk and traffic rules to identify high-risk zones and generate feasible, interpretable avoidance behaviors. A hierarchical warning trigger strategy and lightweight system design enhance runtime efficiency while ensuring real-time responsiveness. Evaluations across four representative high-risk scenarios including car-following braking, cut-in, rear-approaching, and intersection conflict demonstrate REACT's capability to accurately identify critical risks and execute proactive avoidance. Its risk estimation aligns closely with human driver cognition (i.e., warning lead time < 0.4 s), achieving 100% safe avoidance with zero false alarms or missed detections. Furthermore, it exhibits superior real-time performance (< 50 ms latency), strong foresight, and generalization. The lightweight architecture achieves state-of-the-art accuracy, highlighting its potential for real-time deployment in safety-critical autonomous systems.

cs.RO↗

arXiv2025

On Path to Multimodal Generalist: General-Level and General-Bench

The Multimodal Large Language Model (MLLM) is currently experiencing rapid growth, driven by the advanced capabilities of LLMs. Unlike earlier specialists, existing MLLMs are evolving towards a Multimodal Generalist paradigm. Initially limited to understanding multiple modalities, these models have advanced to not only comprehend but also generate across modalities. Their capabilities have expanded from coarse-grained to fine-grained multimodal understanding and from supporting limited modalities to arbitrary ones. While many benchmarks exist to assess MLLMs, a critical question arises: Can we simply assume that higher performance across tasks indicates a stronger MLLM capability, bringing us closer to human-level AI? We argue that the answer is not as straightforward as it seems. This project introduces General-Level, an evaluation framework that defines 5-scale levels of MLLM performance and generality, offering a methodology to compare MLLMs and gauge the progress of existing systems towards more robust multimodal generalists and, ultimately, towards AGI. At the core of the framework is the concept of Synergy, which measures whether models maintain consistent capabilities across comprehension and generation, and across multiple modalities. To support this evaluation, we present General-Bench, which encompasses a broader spectrum of skills, modalities, formats, and capabilities, including over 700 tasks and 325,800 instances. The evaluation results that involve over 100 existing state-of-the-art MLLMs uncover the capability rankings of generalists, highlighting the challenges in reaching genuine AI. We expect this project to pave the way for future research on next-generation multimodal foundation models, providing a robust infrastructure to accelerate the realization of AGI. Project page: https://generalist.top/

cs.CV↗