arXiv subjects

Runze Li

Publications and source records attributed to Runze Li.

At least 37 records · Page 2Linked to original sources

arXiv2026

Mitigating the Curse of Dimensionality in Uniform Convergence of Deep Neural Networks via Smooth Activations

This paper establishes a theoretical framework for the uniform convergence of smoothly activated deep neural network (DNN) estimators. While standard ReLU networks achieve minimax-optimal rates in the $L^2(P)$ norm for various nonparametric regression tasks, we establish a theoretical lower bound demonstrating that least-squares ReLU estimators can suffer from the curse of dimensionality in their uniform convergence behavior. Motivated by the need for reliable uniform guarantees in downstream tasks requiring worst-case reliability, we address this limitation by analyzing smoothly activated DNNs (smooth DNNs), encompassing both feedforward and residual structures. We establish novel pseudo-dimension bounds, non-asymptotic approximation guarantees, and Hölder-norm bounds for the approximators of these models. Leveraging these results, we derive non-asymptotic uniform convergence rates for smooth DNN estimators across multiple statistical contexts, including Huber, least-squares, quantile, and logistic regression. We prove that smooth DNNs can mitigate the {curse of dimensionality} in uniform convergence by adaptively exploiting the low-dimensional hierarchical composition structure of the target function. Supported by both simulation studies and a real-world application, our results position smooth DNNs as a theoretically grounded and practically viable alternative to ReLU networks for statistical learning tasks requiring uniform guarantees.

cs.LG↗

arXiv2026

New Empirical Process Tools and Their Applications to Robust Deep ReLU Networks and Phase Transitions for Nonparametric Regression

This paper introduces new empirical process tools for analyzing a broad class of statistical learning models under heavy-tailed noise and complex function classes. Our primary contribution is the derivation of two Dudley-type maximal inequalities for expected empirical processes that remove restrictive assumptions such as light tails and uniform boundedness of the function class. These inequalities enlarge the scope of empirical process theory available for statistical learning and nonparametric estimation. Exploiting the new bounds, we establish robustness guarantees for deep ReLU network estimators in Huber and quantile regression. In particular, we prove a unified non-asymptotic sub-Gaussian concentration bound that remains valid even under infinite-variance noise and provide a comprehensive analysis of non-asymptotic robustness for deep Huber estimators across all noise regimes. For deep quantile regression, we provide the first non-asymptotic sub-Gaussian bounds without requiring moment assumptions. As an additional application, our framework yields estimation error bounds for nonparametric least-squares estimators that simultaneously accommodate infinite-variance noise, non-Donsker function classes, and approximation error. Moreover, unlike prior approaches based on specialized multiplier processes, our framework extends to broader empirical risk minimization problems, including the nonparametric generalized linear models and the ``set-structured'' models.

math.ST↗

arXiv2026

StepAudio 2.5 Technical Report

Unified audio-language modeling has emerged as a prominent trend in modern speech systems, promising to bring the reasoning capabilities of large language models to auditory tasks. However, existing unified foundations often struggle to match the depth of specialized systems across automatic speech recognition (ASR), text-to-speech synthesis (TTS), and realtime spoken interaction. Bridging this gap remains an open challenge. This report presents StepAudio 2.5, a unified audio-language foundation model that matches or exceeds specialized systems across all three capabilities. Rather than treating these tasks as architecturally distinct, we operate on the premise that once text and audio share a multimodal representational space, task specialization becomes a matter of operational regimes: data construction, optimization targets, and decoding constraints. Guided by this insight, we advance the post-training paradigm from standard supervised learning to task-tailored Reinforcement Learning from Human Feedback (RLHF), using it as the primary mechanism to define complex optimization targets. We leverage this RLHF-centric alignment, alongside specialized decoding, to shape a shared backbone into three distinct operational modes. Concretely, the ASR branch advances transcription efficiency via verifiable multi-token decoding; the TTS branch achieves controllable, expressive synthesis through preference-based RLHF and context-rich supervision; and the Realtime branch realizes low-latency, persona-consistent dialogue via generative reward modeling within an RLHF framework. On standard benchmarks, StepAudio 2.5 achieves state-of-the-art results across ASR, TTS, and Realtime, demonstrating that a singular audio-language foundation can successfully internalize the distinct deployment objectives of speech understanding, generation, and live interaction.

eess.AS↗

arXiv2026

Unified Multimodal Visual Tracking with Dual Mixture-of-Experts

Multimodal visual object tracking can be divided into to several kinds of tasks (e.g. RGB and RGB+X tracking), based on the input modality. Existing methods often train separate models for each modality or rely on pretrained models to adapt to new modalities, which limits efficiency, scalability, and usability. Thus, we introduce OneTrackerV2, a unified multi-modal tracking framework that enables end-to-end training for any modality. We propose Meta Merger to embed multi-modal information into a unified space, allowing flexible modality fusion and robustness. We further introduce Dual Mixture-of-Experts (DMoE): T-MoE models spatio-temporal relations for tracking, while M-MoE embeds multi-modal knowledge, disentangling cross-modal dependencies and reducing feature conflicts. With a shared architecture, unified parameters, and a single end-to-end training, OneTrackerV2 achieves state-of-the-art performance across five RGB and RGB+X tracking tasks and 12 benchmarks, while maintaining high inference efficiency. Notably, even after model compression, OneTrackerV2 retains strong performance. Moreover, OneTrackerV2 demonstrates remarkable robustness under modality-missing scenarios.

cs.CV↗

arXiv2026

World-Value-Action Model: Implicit Planning for Vision-Language-Action Systems

Vision-Language-Action (VLA) models have emerged as a promising paradigm for building embodied agents that ground perception and language into action. However, most existing approaches rely on direct action prediction, lacking the ability to reason over long-horizon trajectories and evaluate their consequences, which limits performance in complex decision-making tasks. In this work, we introduce World-Value-Action (WAV) model, a unified framework that enables implicit planning in VLA systems. Rather than performing explicit trajectory optimization, WAV model learn a structured latent representation of future trajectories conditioned on visual observations and language instructions. A learned world model predicts future states, while a trajectory value function evaluates their long-horizon utility. Action generation is then formulated as inference in this latent space, where the model progressively concentrates probability mass on high-value and dynamically feasible trajectories. We provide a theoretical perspective showing that planning directly in action space suffers from an exponential decay in the probability of feasible trajectories as the horizon increases. In contrast, latent-space inference reshapes the search distribution toward feasible regions, enabling efficient long-horizon decision making. Extensive simulations and real-world experiments demonstrate that the WAV model consistently outperforms state-of-the-art methods, achieving significant improvements in task success rate, generalization ability, and robustness, especially in long-horizon and compositional scenarios. Code is available at https://github.com/Win-commit/WAV.

cs.RO↗

arXiv2026

The Second Challenge on Cross-Domain Few-Shot Object Detection at NTIRE 2026: Methods and Results

Cross-domain few-shot object detection (CD-FSOD) remains a challenging problem for existing object detectors and few-shot learning approaches, particularly when generalizing across distinct domains. As part of NTIRE 2026, we hosted the second CD-FSOD Challenge to systematically evaluate and promote progress in detecting objects in unseen target domains under limited annotation conditions. The challenge received strong community interest, with 128 registered participants and a total of 696 submissions. Among them, 31 teams actively participated, and 19 teams submitted valid final results. Participants explored a wide range of strategies, introducing innovative methods that push the performance frontier under both open-source and closed-source tracks. This report presents a detailed overview of the NTIRE 2026 CD-FSOD Challenge, including a summary of the submitted approaches and an analysis of the final results across all participating teams. Challenge Codes: https://github.com/ohMargin/NTIRE2026_CDFSOD.

cs.CV↗

arXiv2026

EchoTrail-GUI: Building Actionable Memory for GUI Agents via Critic-Guided Self-Exploration

Contemporary GUI agents, while increasingly capable due to advances in Large Vision-Language Models (VLMs), often operate with a critical limitation: they treat each task in isolation, lacking a mechanism to systematically learn from past successes. This digital ''amnesia'' results in sub-optimal performance, repeated errors, and poor generalization to novel challenges. To bridge this gap, we introduce EchoTrail-GUI, a novel framework designed to mimic human-like experiential learning by equipping agents with a dynamic, accessible memory. Our framework operates in three distinct stages. First, during Experience Exploration, an agent autonomously interacts with GUI environments to build a curated database of successful task trajectories, validated by a reward model. Crucially, the entire knowledge base construction is thus fully automated, requiring no human supervision. Second, in the Memory Injection stage, upon receiving a new task, our system efficiently retrieves the most relevant past trajectories to serve as actionable ''memories''. Finally, during GUI Task Inference, these memories are injected as in-context guidance to inform the agent's reasoning and decision-making process. We demonstrate the efficacy of our approach on benchmarks including Android World and AndroidLab. The results show that EchoTrail-GUI significantly improves the task success rate and operational efficiency of baseline agents, validating the power of structured memory in creating more robust and intelligent GUI automation.

cs.AI↗

arXiv2026

Maximum-of-Differences Test for Comparing Multivariate K-Sample Distributions

Comparing $K$-sample distributions is a fundamental problem in data science that arises in a wide variety of fields and applications. In this article, we introduce a maximum-of-differences approach to make such comparisons. Specifically, we first calculate the pairwise distances from the pooled observations of the $K$ samples. We then define the two observations as connected if their distance is less than a pre-specified threshold value. For each observation, we next calculate the ``within" and the ``between" probabilities associated with these two types of connections for the given observation, i.e., with other observations within the same sample and between the given observation and the observations in other samples. Subsequently, we propose a maximum-of-differences (MOD) test that finds the maximum value among the standardized squared differences between the ``within" and the ``between" probabilities of all observations. Accordingly, the proposed test is not only applicable to multivariate data with $K$ samples, but can also be extended to multivariate regression models. Furthermore, we obtain the covariance-adjusted (CA) version of the MOD (CA-MOD) test, which converges to the Type I extreme value distribution under some conditions. Moreover, we demonstrate the asymptotic properties of the two tests under both the null and alternative hypotheses. The performance and usefulness of the tests are illustrated via simulation studies and real examples.

stat.ME↗

arXiv2026

GUIDE: Interpretable GUI Agent Evaluation via Hierarchical Diagnosis

Evaluating GUI agents presents a distinct challenge: trajectories are long, visually grounded, and open-ended, yet evaluation must be both accurate and interpretable. Existing approaches typically apply a single holistic judgment over the entire action-observation sequence-a strategy that proves unreliable on long-horizon tasks and yields binary verdicts offering no insight into where or why an agent fails. This opacity limits the utility of evaluation as a diagnostic tool for agent development. We introduce GUIDE (GUI Understanding and Interpretable Diagnostic Evaluation), a framework that decomposes trajectory assessment into three sequential stages mirroring the compositional structure of GUI tasks. Trajectory Segmentation partitions the full trace into semantically coherent subtask units. Subtask Diagnosis evaluates each unit in context, assigning a completion verdict and generating a structured error analysis with corrective recommendations. Overall Summary aggregates per-subtask diagnoses into a task-level judgment. By operating on bounded subtask segments rather than full trajectories, GUIDE mitigates the context overload that degrades existing evaluators as task complexity grows. We validate GUIDE on three benchmarks: an industrial e-commerce dataset of 932 trajectories, AGENTREWARDBENCH spanning five web agent tasks with 1302 trajectories, and AndroidBench for mobile device control. Across all settings, GUIDE substantially outperforms existing evaluators-achieving up to 5.35 percentage points higher accuracy than the strongest baseline-while producing structured diagnostic reports that directly inform agent improvement.

cs.AI↗

arXiv2026

Stochastic versus Deterministic in Stochastic Gradient Descent

This paper theoretically reanalyzes the convergence of the mini-batch stochastic gradient descent (SGD) for a structured minimization problem involving a finite-sum function with its gradient being stochastically approximated, and an independent term with its gradient being deterministically computed. Rather than collapsing this problem into a standard finite-sum formulation and treating all components uniformly, we study it from a stochastic versus deterministic viewpoint and focus on how these two gradient computations affect mini-batch SGD differently. The step size, the convergence rate, and the radius of the convergence region depend asymmetrically on the characteristics of the two components, which shows the distinct impacts of stochastic approximation versus deterministic computation in the mini-batch SGD. Based on this, we show that our analysis yields a faster convergence rate and a smaller radius of the convergence region. Moreover, an even better convergence rate can be obtained when the independent term endows the objective function with sufficient strong convexity. Also, the convergence rate of our algorithm in expectation approaches that of the classic gradient descent when the batch size increases. Numerical experiments are conducted to support the theoretical analysis as well.

math.OC↗

arXiv2026

STAR: Mitigating Cascading Errors in Spatial Reasoning via Turn-point Alignment and Segment-level DPO

Structured spatial navigation is a core benchmark for Large Language Models (LLMs) spatial reasoning. Existing paradigms like Visualization-of-Thought (VoT) are prone to cascading errors in complex topologies. To solve this, we propose STAR, a two-stage framework grounded on topological anchors, and introduce the RedMaze-23K dataset with human-inspired turnpoint annotations. The first stage uses supervised fine-tuning to help models internalize spatial semantics and prune redundant paths. The second adopts Spatial-aware Segment-level Direct Preference Optimization (SDPO) to refine self-correction in long-horizon navigation. Experiments show STAR achieves state-of-the-art performance among open-source models: its 32B variant outperforms DeepSeek-V3 (29.27% vs. 25.00%) and reaches 82.4% of GPT-4's performance.

cs.CV↗

arXiv2026

Signal-Aware Contrastive Latent Spaces for Anomaly Detection

High-dimensional feature spaces in particle physics events pose a fundamental challenge to density-estimation-based weakly supervised anomaly detection, whose fidelity degrades rapidly with an increasing number of dimensions. We propose a signal-aware latent space construction using supervised contrastive learning trained on simulated Standard Model backgrounds and a diverse set of hypothesized Beyond the Standard Model (BSM) signals. The resulting latent space is low-dimensional, regularized, and signal-sensitive, enabling high-fidelity density estimation for downstream weakly supervised anomaly detection. We demonstrate the approach in a diphoton final state, testing sensitivity across a broad range of BSM scenarios including supersymmetry models, extended Higgs sectors, heavy neutral resonances, and flavor-changing neutral currents. For signals represented in the contrastive training data, the method can elevate discovery sensitivity from previously inaccessible levels to the discovery regime. Critically, the approach retains sensitivity to BSM models not present during training: interpolation and extrapolation to unseen signal topologies yield substantial improvements in expected significance compared to a background-only baseline. By bridging supervised latent space embedding with weakly supervised anomaly detection, this strategy offers a viable path toward anomaly detection in high-dimensional feature spaces at the LHC and beyond.

hep-ph↗

arXiv2026

MERIT: Memory-Enhanced Retrieval for Interpretable Knowledge Tracing

Knowledge Tracing (KT) models students' evolving knowledge states to predict future performance, serving as a foundation for personalized education. While traditional deep learning models achieve high accuracy, they often lack interpretability. Large Language Models (LLMs) offer strong reasoning capabilities but struggle with limited context windows and hallucinations. Furthermore, existing LLM-based methods typically require expensive fine-tuning, limiting scalability and adaptability to new data. We propose MERIT (Memory-Enhanced Retrieval for Interpretable Knowledge Tracing), a training-free framework combining frozen LLM reasoning with structured pedagogical memory. Rather than updating parameters, MERIT transforms raw interaction logs into an interpretable memory bank. The framework uses semantic denoising to categorize students into latent cognitive schemas and constructs a paradigm bank where representative error patterns are analyzed offline to generate explicit Chain-of-Thought (CoT) rationales. During inference, a hierarchical routing mechanism retrieves relevant contexts, while a logic-augmented module applies semantic constraints to calibrate predictions. By grounding the LLM in interpretable memory, MERIT achieves state-of-the-art performance on real-world datasets without gradient updates. This approach reduces computational costs and supports dynamic knowledge updates, improving the accessibility and transparency of educational diagnosis.

cs.CL↗

arXiv2026

CRL-VLA: Continual Vision-Language-Action Learning

Lifelong learning is critical for embodied agents in open-world environments, where reinforcement learning fine-tuning has emerged as an important paradigm to enable Vision-Language-Action (VLA) models to master dexterous manipulation through environmental interaction. Thus, Continual Reinforcement Learning (CRL) is a promising pathway for deploying VLA models in lifelong robotic scenarios, yet balancing stability (retaining old skills) and plasticity (learning new ones) remains a formidable challenge for existing methods. We introduce CRL-VLA, a framework for continual post-training of VLA models with rigorous theoretical bounds. We derive a unified performance bound linking the stability-plasticity trade-off to goal-conditioned advantage magnitude, scaled by policy divergence. CRL-VLA resolves this dilemma via asymmetric regulation: constraining advantage magnitudes on prior tasks while enabling controlled growth on new tasks. This is realized through a simple but effective dual-critic architecture with novel Goal-Conditioned Value Formulation (GCVF), where a frozen critic anchors semantic consistency and a trainable estimator drives adaptation. Experiments on the LIBERO benchmark demonstrate that CRL-VLA effectively harmonizes these conflicting objectives, outperforming baselines in both anti-forgetting and forward adaptation.

cs.AI↗

arXiv2026

GR3EN: Generative Relighting for 3D Environments

We present a method for relighting 3D reconstructions of large room-scale environments. Existing solutions for 3D scene relighting often require solving under-determined or ill-conditioned inverse rendering problems, and are as such unable to produce high-quality results on complex real-world scenes. Though recent progress in using generative image and video diffusion models for relighting has been promising, these techniques are either limited to 2D image and video relighting or 3D relighting of individual objects. Our approach enables controllable 3D relighting of room-scale scenes by distilling the outputs of a video-to-video relighting diffusion model into a 3D reconstruction. This side-steps the need to solve a difficult inverse rendering problem, and results in a flexible system that can relight 3D reconstructions of complex real-world scenes. We validate our approach on both synthetic and real-world datasets to show that it can faithfully render novel views of scenes under new lighting conditions.

cs.CV↗

arXiv2026

Uncertainty-Aware Data-Based Method for Fast and Reliable Shape Optimization

Data-based optimization (DBO) offers a promising approach for efficiently optimizing shape for better aerodynamic performance by leveraging a pretrained surrogate model for offline evaluations during iterations. However, DBO heavily relies on the quality of the training database. Samples outside the training distribution encountered during optimization can lead to significant prediction errors, potentially misleading the optimization process. Therefore, incorporating uncertainty quantification into optimization is critical for detecting outliers and enhancing robustness. This study proposes an uncertainty-aware data-based optimization (UA-DBO) framework to monitor and minimize surrogate model uncertainty during DBO. A probabilistic encoder-decoder surrogate model is developed to predict uncertainties associated with its outputs, and these uncertainties are integrated into a model-confidence-aware objective function to penalize samples with large prediction errors during data-based optimization process. The UA-DBO framework is evaluated on two multipoint optimization problems aimed at improving airfoil drag divergence and buffet performance. Results demonstrate that UA-DBO consistently reduces prediction errors in optimized samples and achieves superior performance gains compared to original DBO. Moreover, compared to multipoint optimization based on full computational simulations, UA-DBO offers comparable optimization effectiveness while significantly accelerating optimization speed.

cs.LG↗

arXiv2026

A Comprehensive Exploration of Personalized Learning in Smart Education: From Student Modeling to Personalized Recommendations

With the development of artificial intelligence, personalized learning has attracted much attention as an integral part of intelligent education. In recent years, countries and regions such as China, the United States, and the European Union have increasingly recognized the importance of personalized learning, emphasizing its potential to integrate large-scale education with individualized instruction effectively. This survey provides a comprehensive analysis of personalized learning by reviewing relevant studies published in major conferences and journals between January 2017 and April 2025. We examine its definition, objectives, and underlying educational theories, highlighting its pedagogical significance. Furthermore, we explore personalized learning from two key dimensions: student modeling and personalized recommendations. Student modeling is analyzed from both cognitive and non-cognitive perspectives, while recommendation approaches are categorized based on their specific objectives. Additionally, we investigate the interplay between these components and their role in enhancing personalized learning. Beyond theoretical and algorithmic insights, this survey reviews real-world applications, demonstrating personalized learning's effectiveness in educational practice. Finally, we discuss key challenges and future directions, offering a multidimensional perspective that bridges theory and practice.

cs.CY↗

arXiv2026

Measuring Love Toward AI: Development and Validation of the Love Attitudes Scale toward Artificial Intelligence (LAS-AI)

Artificial intelligences (AIs) are increasingly capable of emotionally engaging with humans to the point of forming intimate relationships. Yet, current studies on romantic love toward AI lack statistically validated instruments to measure romantic love toward AI, hindering empirical research. To address this gap, we reinterpreted Lee's love styles theory in the AI context and developed the Love Attitudes Scale toward AI (LAS-AI). The resulting 24-item, six-factor scale was validated across four phases using three independent samples (N = 899), demonstrating strong psychometric properties. The findings further revealed that people primarily seek practical, passionate, and companionship-based relationships with AI (i.e., Pragma, Eros, and Storge), showing little interest in a playful or noncommittal approach (i.e., Ludus). We also provided an initial exploration of the similarities and differences between romantic love with humans and AI. The LAS-AI offers a robust tool for future research on human-AI romantic relationships, with prolific implications.

cs.HC↗