arXiv subjects

Xinyu Zhao

Publications and source records attributed to Xinyu Zhao.

At least 37 records · Page 2Linked to original sources

arXiv2026

Reinforcement Learning for Fast and Robust Longitudinal Qubit Readout

Longitudinal coupling offers a compelling pathway for quantum nondemolition (QND) readout, but pulse design is constrained by hardware limitations such as the coupling strength and the photon number required to stay within the linear regime. We develop a reinforcement learning framework to optimize the longitudinal coupling waveform under such constraints. Building upon the theoretical foundation of shortcuts to adiabaticity (STA), we parameterize an auxiliary trajectory with cubic B-splines and reconstruct the physical control. At a fixed short readout time, the optimized pulse converges to a constraint saturating flat-top protocol and yields a approximately $50\%$ improvement in $\mathrm{SNR}$ over an STA baseline, while exhibiting enhanced robustness to parameter drifts. Simulation results demonstrate the efficacy of reinforcement learning in optimizing longitudinal readout pulses. The optimized protocol attains substantial performance gains and yields smooth, hardware-compatible waveforms governed by an interpretable ``saturate-and-hold'' mechanism.

quant-ph↗

arXiv2026

Clue Matters: Leveraging Latent Visual Clues to Empower Video Reasoning

Multi-modal Large Language Models (MLLMs) have significantly advanced video reasoning, yet Video Question Answering (VideoQA) remains challenging due to its demand for temporal causal reasoning and evidence-grounded answer generation. Prevailing end-to-end MLLM frameworks lack explicit structured reasoning between visual perception and answer derivation, causing severe hallucinations and poor interpretability. Existing methods also fail to address three core gaps: faithful visual clue extraction, utility-aware clue filtering, and end-to-end clue-answer alignment. Inspired by hierarchical human visual cognition, we propose ClueNet, a clue-aware video reasoning framework with a two-stage supervised fine-tuning paradigm without extensive base model modifications. Decoupled supervision aligns clue extraction and chain-based reasoning, while inference supervision with an adaptive clue filter refines high-order reasoning, alongside lightweight modules for efficient inference. Experiments on NExT-QA, STAR, and MVBench show that ClueNet outperforms state-of-the-art methods by $\ge$ 1.1%, with superior generalization, hallucination mitigation, inference efficiency, and cross-backbone compatibility. This work bridges the perception-to-generation gap in MLLM video understanding, providing an interpretable, faithful reasoning paradigm for high-stakes VideoQA applications.

cs.CV↗

arXiv2026

FaithCoT-Bench: Benchmarking Instance-Level Faithfulness of Chain-of-Thought Reasoning

Large language models (LLMs) increasingly rely on Chain-of-Thought (CoT) prompting to improve problem-solving and provide seemingly transparent explanations. However, growing evidence shows that CoT often fail to faithfully represent the underlying reasoning process, raising concerns about their reliability in high-risk applications. Although prior studies have focused on mechanism-level analyses showing that CoTs can be unfaithful, they leave open the practical challenge of deciding whether a specific trajectory is faithful to the internal reasoning of the model. To address this gap, we introduce FaithCoT-Bench, a unified benchmark for instance-level CoT unfaithfulness detection. Our framework establishes a rigorous task formulation that formulates unfaithfulness detection as a discriminative decision problem, and provides FINE-CoT (Faithfulness instance evaluation for Chain-of-Thought), an expert-annotated collection of over 1,000 trajectories generated by four representative LLMs across four domains, including more than 300 unfaithful instances with fine-grained causes and step-level evidence. We further conduct a systematic evaluation of eleven representative detection methods spanning counterfactual, logit-based, and LLM-as-judge paradigms, deriving empirical insights that clarify the strengths and weaknesses of existing approaches and reveal the increased challenges of detection in knowledge-intensive domains and with more advanced models. To the best of our knowledge, FaithCoT-Bench establishes the first comprehensive benchmark for instance-level CoT faithfulness, setting a solid basis for future research toward more interpretable and trustworthy reasoning in LLMs.

cs.AI↗

arXiv2026

Dialogue is Better Than Monologue: Instructing Medical LLMs via Strategical Conversations

Current medical AI systems often fail to replicate real-world clinical reasoning, as they are predominantly trained and evaluated on static text and question-answer tasks. These tuning methods and benchmarks overlook critical aspects like evidence-based reasoning and handling distracting information. To bridge this gap, we introduce a novel benchmark that simulates real-world diagnostic scenarios, integrating noise and difficulty levels aligned with USMLE standards. Moreover, we explore dialogue-based fine-tuning, which transforms static datasets into conversational formats to better capture iterative reasoning processes. Experiments show that dialogue-tuned models outperform traditional methods, with improvements of $9.64\%$ in multi-round reasoning scenarios and $6.18\%$ in accuracy in a noisy environment. Our findings highlight dialogue tuning as a promising approach for advancing clinically aligned and robust medical AI systems.

cs.CL↗

arXiv2026

Giant atoms coupled to waveguide: Continuous coupling and multiple excitations

We propose a stochastic Schrödinger equation (SSE) approach to investigate the dynamics of giant atoms coupled to a waveguide, addressing two critical gaps in existing research, namely insufficient exploration on continuous coupling and multiple excitations. A key finding is that continuous coupling, unlike discrete coupling at finite points, breaks the constant phase difference condition, thereby weakening the interference effects in giant atom-waveguide systems. In addition, a key technical advantage of the SSE approach is that auto- and cross-correlation functions can directly reflect the complex photon emission/absorption processes and time-delay effects in giant atom-waveguide systems. Moreover, the SSE approach also naturally handles multiple excitations, without increasing equation complexity as the number of excitations grows. This feature enables the investigation of multi-excitation initial states of the waveguide, such as thermal and squeezed initial states. Overall, our approach provides a powerful tool for studying the dynamics of giant atoms coupled to waveguide, particularly for continuous coupling and multi-excitation systems.

quant-ph↗

arXiv2026

Non-Markovian environment induced chaos in optomechanical system

In traditional research, chaos is frequently accompanied by non-linearity, which typically stems from non-linear interactions or external driving forces. However, in this paper, we present the chaotic behavior that is completely attributed to the non-linear back-reaction of non-Markovian environment. To be specific, we derive the dynamical equations of an optomechanical system and demonstrate that the non-linearity (cause of chaos) in the equations arises entirely from the time-domain convolutions (TDCs) induced by non-Markovian corrections. Under Markovian conditions, these TDCs are reduced into constants, thereby losing the nonlinearity and ultimately leading to the disappearance of chaos. Furthermore, we also observe chaos generation in the absence of optomechanical couplings, which further confirms that the non-Markovian effect is the sole inducement of chaos and the environmental parameters play important roles in the generation of chaos. We hope these results may open a new direction to investigate chaotic dynamics purely caused by non-Markovian environments.

quant-ph↗

arXiv2026

Non-Markovian environment induced Schrödinger cat state transfer in an optical Newton's cradle

In this manuscript, we study the Schrödinger cat state transfer in a quantum optical version of Newton's cradle in non-Markovian environment. Based on a non-Markovian master equation, we show that the cat state can be transferred purely through the memory effect of the non-Markovian common environment, even without any direct couplings between neighbor cavities. The mechanism of the environment induced cat state transfer is analyzed both analytically and numerically to demonstrate that the transfer is a unique phenomenon in non-Markovian regime. From this example, the non-Markovian environment is shown to be qualitatively different from the Markovian environment reflected by the finite versus zero residue coherence. Besides, we also show the influence of environmental parameters are crucial for the transfer. We hope the cat state transfer studied in this work may shed more light on the fundamental difference between non-Markovian and Markovian environments.

quant-ph↗

arXiv2026

ChartMuseum: Testing Visual Reasoning Capabilities of Large Vision-Language Models

Chart understanding presents a unique challenge for large vision-language models (LVLMs), as it requires the integration of sophisticated textual and visual reasoning capabilities. However, current LVLMs exhibit a notable imbalance between these skills, falling short on visual reasoning that is difficult to perform in text. We conduct a case study using a synthetic dataset solvable only through visual reasoning and show that model performance degrades significantly with increasing visual complexity, while human performance remains robust. We then introduce ChartMuseum, a new Chart Question Answering (QA) benchmark containing 1,162 expert-annotated questions spanning multiple reasoning types, curated from real-world charts across 184 sources, specifically built to evaluate complex visual and textual reasoning. Unlike prior chart understanding benchmarks -- where frontier models perform similarly and near saturation -- our benchmark exposes a substantial gap between model and human performance, while effectively differentiating model capabilities: although humans achieve 93% accuracy, the best-performing model Gemini-2.5-Pro attains only 63.0%, and the leading open-source LVLM Qwen2.5-VL-72B-Instruct achieves only 38.5%. Moreover, on questions requiring primarily visual reasoning, all models experience a 35%-55% performance drop from text-reasoning-heavy question performance. Lastly, our qualitative error analysis reveals specific categories of visual reasoning that are challenging for current LVLMs.

cs.CL↗

arXiv2026

Towards Stabilized and Efficient Diffusion Transformers through Long-Skip-Connections with Spectral Constraints

Diffusion Transformers (DiT) have emerged as a powerful architecture for image and video generation, offering superior quality and scalability. However, their practical application suffers from inherent dynamic feature instability, leading to error amplification during cached inference. Through systematic analysis, we identify the absence of long-range feature preservation mechanisms as the root cause of unstable feature propagation and perturbation sensitivity. To this end, we propose Skip-DiT, an image and video generative DiT variant enhanced with Long-Skip-Connections (LSCs) - the key efficiency component in U-Nets. Theoretical spectral norm and visualization analysis demonstrate how LSCs stabilize feature dynamics. Skip-DiT architecture and its stabilized dynamic feature enable an efficient statical caching mechanism that reuses deep features across timesteps while updating shallow components. Extensive experiments across the image and video generation tasks demonstrate that Skip-DiT achieves: (1) 4.4 times training acceleration and faster convergence, (2) 1.5-2 times inference acceleration with negligible quality loss and high fidelity to the original output, outperforming existing DiT caching methods across various quantitative metrics. Our findings establish Long-Skip-Connections as critical architectural components for stable and efficient diffusion transformers. Codes are provided in the https://github.com/OpenSparseLLMs/Skip-DiT.

cs.CV↗

arXiv2026

Two Heads are Better than One: Distilling Large Language Model Features Into Small Models with Feature Decomposition and Mixture

Market making (MM) through Reinforcement Learning (RL) has attracted significant attention in financial trading. With the development of Large Language Models (LLMs), more and more attempts are being made to apply LLMs to financial areas. A simple, direct application of LLM as an agent shows significant performance. Such methods are hindered by their slow inference speed, while most of the current research has not studied LLM distillation for this specific task. To address this, we first propose the normalized fluorescent probe to study the mechanism of the LLM's feature. Based on the observation found by our investigation, we propose Cooperative Market Making (CMM), a novel framework that decouples LLM features across three orthogonal dimensions: layer, task, and data. Various student models collaboratively learn simple LLM features along with different dimensions, with each model responsible for a distinct feature to achieve knowledge distillation. Furthermore, CMM introduces an Hájek-MoE to integrate the output of the student models by investigating the contribution of different models in a kernel function-generated common feature space. Extensive experimental results on four real-world market datasets demonstrate the superiority of CMM over the current distillation method and RL-based market-making strategies.

cs.AI↗

arXiv2026

H3PIMAP: A Heterogeneity-Aware Multi-Objective DNN Mapping Framework on Electronic-Photonic Processing-in-Memory Architectures

The future of artificial intelligence (AI) acceleration demands a paradigm shift beyond the limitations of purely electronic or photonic architectures. Photonic analog computing delivers unmatched speed and parallelism but struggles with data movement, robustness, and precision, while electronic processing-in-memory (PIM) enables energy-efficient computing by co-locating storage and computation but suffers from endurance and reconfiguration constraints, limiting it to static weight mapping. Neither approach alone achieves the balance needed for adaptive, efficient AI. To break this impasse, we study a hybrid electronic-photonic-PIM computing architecture and introduce H3PIMAP, a heterogeneity-aware mapping framework that seamlessly orchestrates workloads across electronic and optical tiers. By optimizing workload partitioning through a two-stage multi-objective exploration method, H3PIMAP harnesses light speed for high-throughput operations and PIM efficiency for memory-bound tasks. In system-level evaluations, H3PIMAP delivers a 3.32x latency reduction across language and vision models and, on large language models, achieves 77.0% lower latency with 14.6% lower energy at matched quality, outperforming homogeneous and naive mapping strategies. This proposed framework lays the foundation for hybrid AI accelerators, bridging the gap between electronic and photonic computation for next-generation efficiency and scalability.

cs.AR↗

arXiv2026

Metacognitive Self-Correction for Multi-Agent System via Prototype-Guided Next-Execution Reconstruction

Large Language Model based multi-agent systems (MAS) excel at collaborative problem solving but remain brittle to cascading errors: a single faulty step can propagate across agents and disrupt the trajectory. In this paper, we present MASC, a metacognitive framework that endows MAS with real-time, unsupervised, step-level error detection and self-correction. MASC rethinks detection as history-conditioned anomaly scoring via two complementary designs: (1) Next-Execution Reconstruction, which predicts the embedding of the next step from the query and interaction history to capture causal consistency, and (2) Prototype-Guided Enhancement, which learns a prototype prior over normal-step embeddings and uses it to stabilize reconstruction and anomaly scoring under sparse context (e.g., early steps). When an anomaly step is flagged, MASC triggers a correction agent to revise the acting agent's output before information flows downstream. On the Who&When benchmark, MASC consistently outperforms all baselines, improving step-level error detection by up to 8.47% AUC-ROC ; When plugged into diverse MAS frameworks, it delivers consistent end-to-end gains across architectures, confirming that our metacognitive monitoring and targeted correction can mitigate error propagation with minimal overhead.

cs.AI↗

arXiv2025

Entanglement protection induced by mixed noise

Contrary to the conventional view that noise is detrimental, we show that mixed noise can protect entanglement in a two-atom-cavity system. Specifically, the leakage of the cavity and the stochastic atom-cavity couplings are modeled as two types of noises. From the analytical derivation of the dynamical equations, the mechanism of the entanglement protection is revealed as the high-frequency(HF) noise in the atom-cavity couplings could suppress the decoherence caused by the cavity leakage, thus protect the entanglement. We investigate the entanglement protection induced by mixed noise constructed from diverse noise types, including the Ornstein-Uhlenbeck noise, flicker noise, and telegraph noise. Numerical simulations demonstrate that entanglement protection depends critically on the proportion of HF components in the power spectral density of the mixed noise. Our work establishes that enhanced HF components are essential for effective noise-assisted entanglement protection, offering key insights for noise engineering in practical open quantum systems.

quant-ph↗

arXiv2025

Partially Observable Markov Decision Process Framework for Operating Condition Optimization Using Real-Time Degradation Signals

In many engineering systems, proper predictive maintenance and operational control are essential to increase efficiency and reliability while reducing maintenance costs. However, one of the major challenges is that many sensors are used for system monitoring. Analyzing these sensors simultaneously for better predictive maintenance optimization is often very challenging. In this paper, we propose a systematic decision-making framework to improve the system performance in manufacturing practice, considering the real-time degradation signals generated by multiple sensors. Specifically, we propose a partially observed Markov decision process (POMDP) model to generate the optimal capacity and predictive maintenance policies, given the fact that the observation of the system state is imperfect. Such work provides a systematic approach that focuses on jointly controlling the operating conditions and preventive maintenance utilizing the real-time machine deterioration signals by incorporating the degradation constraint and non-observable states. We apply this technique to the bearing degradation data and NASA aircraft turbofan engine dataset, demonstrating the effectiveness of the proposed method.

stat.AP↗

arXiv2025

Vulnerability-Aware Robust Multimodal Adversarial Training

Multimodal learning has shown significant superiority on various tasks by integrating multiple modalities. However, the interdependencies among modalities increase the susceptibility of multimodal models to adversarial attacks. Existing methods mainly focus on attacks on specific modalities or indiscriminately attack all modalities. In this paper, we find that these approaches ignore the differences between modalities in their contribution to final robustness, resulting in suboptimal robustness performance. To bridge this gap, we introduce Vulnerability-Aware Robust Multimodal Adversarial Training (VARMAT), a probe-in-training adversarial training method that improves multimodal robustness by identifying the vulnerability of each modality. To be specific, VARMAT first explicitly quantifies the vulnerability of each modality, grounded in a first-order approximation of the attack objective (Probe). Then, we propose a targeted regularization term that penalizes modalities with high vulnerability, guiding robust learning while maintaining task accuracy (Training). We demonstrate the enhanced robustness of our method across multiple multimodal datasets involving diverse modalities. Finally, we achieve {12.73%, 22.21%, 11.19%} robustness improvement on three multimodal datasets, revealing a significant blind spot in multimodal adversarial training.

cs.LG↗

arXiv2025

SARHAchat: An LLM-Based Chatbot for Sexual and Reproductive Health Counseling

While Artificial Intelligence (AI) shows promise in healthcare applications, existing conversational systems often falter in complex and sensitive medical domains such as Sexual and Reproductive Health (SRH). These systems frequently struggle with hallucination and lack the specialized knowledge required, particularly for sensitive SRH topics. Furthermore, current AI approaches in healthcare tend to prioritize diagnostic capabilities over comprehensive patient care and education. Addressing these gaps, this work at the UNC School of Nursing introduces SARHAchat, a proof-of-concept Large Language Model (LLM)-based chatbot. SARHAchat is designed as a reliable, user-centered system integrating medical expertise with empathetic communication to enhance SRH care delivery. Our evaluation demonstrates SARHAchat's ability to provide accurate and contextually appropriate contraceptive counseling while maintaining a natural conversational flow. The demo is available at https://sarhachat.com/}{https://sarhachat.com/.

cs.CY↗

arXiv2025

L2M-AID: Autonomous Cyber-Physical Defense by Fusing Semantic Reasoning of Large Language Models with Multi-Agent Reinforcement Learning (Preprint)

The increasing integration of Industrial IoT (IIoT) exposes critical cyber-physical systems to sophisticated, multi-stage attacks that elude traditional defenses lacking contextual awareness. This paper introduces L2M-AID, a novel framework for Autonomous Industrial Defense using LLM-empowered, Multi-agent reinforcement learning. L2M-AID orchestrates a team of collaborative agents, each driven by a Large Language Model (LLM), to achieve adaptive and resilient security. The core innovation lies in the deep fusion of two AI paradigms: we leverage an LLM as a semantic bridge to translate vast, unstructured telemetry into a rich, contextual state representation, enabling agents to reason about adversary intent rather than merely matching patterns. This semantically-aware state empowers a Multi-Agent Reinforcement Learning (MARL) algorithm, MAPPO, to learn complex cooperative strategies. The MARL reward function is uniquely engineered to balance security objectives (threat neutralization) with operational imperatives, explicitly penalizing actions that disrupt physical process stability. To validate our approach, we conduct extensive experiments on the benchmark SWaT dataset and a novel synthetic dataset generated based on the MITRE ATT&CK for ICS framework. Results demonstrate that L2M-AID significantly outperforms traditional IDS, deep learning anomaly detectors, and single-agent RL baselines across key metrics, achieving a 97.2% detection rate while reducing false positives by over 80% and improving response times by a factor of four. Crucially, it demonstrates superior performance in maintaining physical process stability, presenting a robust new paradigm for securing critical national infrastructure.

cs.AI↗

arXiv2025

Task-Aware Resolution Optimization for Visual Large Language Models

Real-world vision-language applications demand varying levels of perceptual granularity. However, most existing visual large language models (VLLMs), such as LLaVA, pre-assume a fixed resolution for downstream tasks, which leads to subpar performance. To address this problem, we first conduct a comprehensive and pioneering investigation into the resolution preferences of different vision-language tasks, revealing a correlation between resolution preferences with image complexity, and uncertainty variance of the VLLM at different image input resolutions. Building on this insight, we propose an empirical formula to determine the optimal resolution for a given vision-language task, combining these two factors. Second, based on rigorous experiments, we propose a novel parameter-efficient fine-tuning technique to extend the visual input resolution of pre-trained VLLMs to the identified optimal resolution. Extensive experiments on various vision-language tasks validate the effectiveness of our method.

cs.CV↗