Searcharxiv⌕ Search

arXiv subjects

Yuxin Li

Publications and source records attributed to Yuxin Li.

At least 19 recordsLinked to original sources
arXiv2026

ParallelWorld: Test-Time Scaling for Embodied Reasoning

Embodied Reasoning constitutes a fundamental capability of embodied intelligence, serving as the basis for autonomous perception, reasoning, and interaction within physical environments. Recent studies have shifted the paradigm of embodied reasoning from static perception toward dynamic exploration, where agents acquire task-relevant information through interactions with the environment. However, existing active reasoning approaches generally generate exploration trajectories incrementally without long-horizon planning. Even recently emerged test-time scaling frameworks often resort to myopic, single-step lookaheads, which struggle to resolve the delayed feedback inherent in complex, occluded spatial environments. To address this limitation, we propose ParallelWorld, a multi-horizon test-time scaling framework for embodied reasoning. Instead of greedy, single-step trials, ParallelWorld empowers agents to simulate and evaluate multi-step future trajectories in parallel before committing to an action. Specifically, we introduce a verifier-guided tree-search paradigm. Starting from the current state, ParallelWorld branches into multiple parallel trajectories and rolls them out continuously across a multi-step horizon. At each simulation step, a verifier agent evaluates the intermediate state transitions, dynamically pruning unpromising branches and prioritizing paths with the highest information gain. Once the multi-step prospective simulation is complete, the agent synthesizes the long-horizon outcomes to commit to the optimal action sequence. Finally, an answer agent performs reasoning over the selected trajectory to produce the final reasoning. Extensive experiments on ESI-Bench demonstrate that ParallelWorld consistently improves active perception and reasoning performance.

cs.AI↗
arXiv2026

ParaASR: Multi-Token Prediction for Fast and Long-Context LLM-Based Speech Recognition

Audio-encoder-LLM-decoder architectures have become the dominant paradigm for modern automatic speech recognition (ASR), improving transcription quality through large-scale language modeling. However, the cost of autoregressive decoding scales with decoder size, creating a fundamental trade-off between recognition quality and serving latency. We argue this trade-off is not inherent: unlike open-ended text generation, ASR outputs are strongly anchored to the input speech signal, providing a natural inductive bias toward high-parallelism decoding. Building on this, we introduce ParaASR, an ASR system that leverages Multi-Token Prediction (MTP) to let a 4B LLM decoder emit multiple tokens per forward step. Starting from a publicly available audio-language foundation, the model first establishes a robust autoregressive recognizer and then aligns five future-token branches through a staged optimization recipe. At inference, it proposes a six-token continuation per step and admits only the verified prefix into the transcript, preserving the safety of standard autoregressive decoding. The average accepted length reaches 5.0 out of 6 proposed tokens, confirming that the deterministic structure of speech makes ASR an especially natural setting for multi-token decoding. ParaASR further retains a native 32K-context window and transcribes up to 30 minutes of audio in a single pass. Across diverse benchmarks, it attains average error rates of 2.97%, 3.68%, and 3.70% on Chinese, English, and long-form evaluations, respectively, while reaching a real-time factor (RTF) as low as 0.0053. These results show that decoder scaling, low-latency inference, and long-context transcription need not be competing goals when future-token proposals are anchored by the acoustic signal and guarded by autoregressive verification.

cs.SD↗
arXiv2026

FDDWAN: A Frequency-Decoupled Diffusion Network for Watermarking Attack

Existing invisible watermark removal methods often struggle to accurately capture the watermark-bearing features, leading to an unfavorable trade-off between watermark suppression and perceptual fidelity. In this paper, we propose the Frequency-Decoupled Diffusion Watermark Attack Network (FDDWAN), a coarse-to-fine framework that performs watermark removal through wavelet-domain decomposition and residual diffusion refinement. In the initial stage, the Wavelet-based Frequency-domain Preliminary Attack Module (WFPAM) decomposes the watermarked image into low- and high-frequency subbands and applies frequency-specific attack strategies tailored to their respective contributions to watermark robustness and perceptual quality. In the next stage, the Frequency-domain Residual Diffusion Attack Module (FRDAM) separately models the residual distributions between the preliminarily attacked outputs and the corresponding watermark-free references during training. Rather than reconstructing the entire image, FRDAM selectively refines frequency-domain residuals, directing the diffusion process toward the remaining watermark related discrepancies while minimizing modifications to image content. Extensive experiments on CelebA and ImageNet across four representative watermarking schemes demonstrate that FDDWAN achieves a more favorable trade-off between watermark removal effectiveness and visual fidelity than conventional and learning-based attack methods.

cs.CV↗
arXiv2026

Normalized solutions for a class of gradient-type Schr\"odinger systems under Neumann boundary conditions in bounded domains

We investigate the existence of normalized solutions to the gradient-type Schr\"odinger system \begin{equation*} \begin{cases} -\Delta u+ V_1(x)u+\lambda u= uv^2 & \text{ in } \Omega,\\ -\Delta v+ V_2(x)v+\lambda v= u^2v & \text{ in } \Omega %\frac{\partial u}{\partial \nu}=\frac{\partial v}{\partial \nu}=0 \, & \text{ on } \partial \Omega \end{cases} \end{equation*} subject to the mass constraint $\int_{\Omega}\left(|u|^2+|v|^2 \right)dx=a>0$ and Neumann boundary conditions, where $\Omega\subset \mathbb{R}^3$ is a smooth bounded domain, each $V_i$ is continuous, and $\lambda$ is a Lagrange multiplier. Applying a minimax principle that incorporates Morse index information, we establish the existence of nontrivial normalized solutions of mountain pass type. The proof is based on a refined blow-up analysis adapted to such gradient-type systems, together with new Liouville-type theorems for finite Morse index solutions of the associated limit systems in $\mathbb{R}^3$ and $\mathbb{R}^3_+$.

math.AP↗
arXiv2026

Index SLM Technical Report

We present Index-1.9B, a series of open small language models developed at Bilibili. The series comprises four models: Index-1.9B-Base, a foundation model with 1.9 billion non-embedding parameters pre-trained on 2.8 trillion predominantly Chinese and English tokens; Index-1.9B-Pure, a control variant trained with an identical recipe but with all instruction-like data strictly filtered from the corpus; Index-1.9B-Chat, aligned from the base model with supervised fine-tuning and direct preference optimization; and Index-1.9B-Character, which augments the chat model with retrieval-augmented generation for few-shot role-playing customization. Pre-training employs a Warmup-Stable-Decay learning-rate schedule in which the concentration of curated data is raised substantially during the decay phase, together with a Norm-Head output layer that stabilizes training under large learning rates. On a suite of standard benchmarks covering examination, reasoning, mathematics, and code, Index-1.9B-Base attains an average score of 64.92, competitive with or exceeding open models of several times its size. We further report controlled studies on model depth, learning-rate magnitude and scheduling, the interaction between learning-rate decay and data quality, and the effect of including instruction data during pre-training, and we document an unexplained surge in benchmark performance midway through the constant-learning-rate phase. All models, together with evaluation code, are released at https://github.com/bilibili/Index-1.9B.

cs.CL↗
arXiv2026

Decoupling Conversational Dynamics in Full-Duplex Spoken Models through Reinforcement Learning

Recent full-duplex spoken dialogue models have demonstrated compelling progress toward human-like interaction, enabling agents to respond with low latency, produce backchannels, and handle user barge-ins. Yet these improvements in conversational dynamics often come with weaker reasoning and instruction-following abilities, revealing a potential tension between interactive dynamics and intelligence capability. In this paper, we argue that such an intelligence--dynamics trade-off is not fundamental: conversational dynamics can instead be learned as a separate real-time decision policy from human dialogue data. To this end, we propose DuplexPO, a reinforcement learning (RL) framework that decouples when to speak from what to say. It preserves the semantic response capability of an instruction-tuned assistant, while optimizing its temporal interaction behavior over selected high-impact windows from long human conversations. To quantitatively optimize these dynamics, we formulate the Factorized Conversational Dynamics Reward (FCDR) to enable fine-grained temporal credit assignment for turn initiation, backchanneling, yielding, and regularized participation. The policy is then optimized with a GRPO-style objective. Experiments show that DuplexPO substantially improves full-duplex behaviors, including timely backchannels, smooth turn-taking, and barge-in handling, while maintaining strong reasoning and instruction-following performance. Moreover, improvements in dynamics-oriented metrics are reflected in better user experience, suggesting that optimizing conversational timing as a standalone objective can promote more natural full-duplex interaction.

eess.AS↗
arXiv2026

A Spectral Solver for Acoustic Scattering by Multiple Quasi-Axisymmetric Structures

Acoustic scattering arises in a wide range of applications, including medical imaging, geophysical exploration, acoustic metamaterials, etc. In this paper, we develop a fast and highly accurate algorithm for acoustic scattering by multiple quasi-axisymmetric objects, whose axis of rotation is an arbitrary curve. The method is based on a Nystr\"om discretization that combines Gauss-Legendre quadrature with the trapezoidal rule. To treat the singular integrals that occur when target points are close to or coincide with source points, we reformulate them as evaluations of the modal Green's function and its derivatives, which are computed efficiently using the fast Fourier transform and convolution. The multiple scattering solver is then constructed by coupling the single scatterer discretizations through inter-body boundary integral interactions. We also present a convergence analysis for scattering problems with smooth geometries. Numerical examples demonstrate the efficiency and accuracy of the proposed method for solving multiple scattering problems involving up to 1000 quasi-axisymmetric structures.

math.NA↗
arXiv2026

MBench: A Comprehensive Benchmark on Memory Capability for Video World Models

Recent advancements in video-based world models have demonstrated an unprecedented ability to synthesize high-fidelity visual sequences. However, a fundamental gap persists between visually plausible video generation and the functional requirements of a world model, particularly in maintaining a stable and reasonable internal state over extended temporal horizons. While existing benchmarks primarily emphasize visual quality, motion coherence, and text-video alignment, they largely overlook memory, the core capability of a world model to preserve consistency across long-term horizons and complex interactions. To address this gap, we present \textbf{MBench}, a comprehensive benchmark dedicated to quantifying and evaluating the memory capability of video world models. We systematically decompose the memory capability of video world models into three hierarchical and complementary core dimensions: entity consistency, environment consistency, and causal consistency, which are further refined into 12 quantifiable sub-dimensions for comprehensive characterization of long-term memory. Our benchmark is built upon rigorously curated real-captured long videos, and evaluated by rule-based quantitative matrices and VLM to enable objective and comprehensive consistency assessment. Extensive evaluations of mainstream state-of-the-art video world models reveal critical systemic limitations of existing methods in long-term state retention, providing a standardized benchmark and clear research direction to advance the field.

cs.CV↗
arXiv2026

The WER Trap: Shattering the Illusion of Unified Tokens in Speech Language Models

The pursuit of a "unified" discrete token for both speech understanding and generation has led the Speech Language Model (SLM) community to heavily rely on Word Error Rate (WER) -- the core metric for Whisper-style tokenizers -- as the definitive proxy for representation quality. This fosters the assumption that low-WER tokens inherently preserve the information necessary for intelligible acoustic synthesis. We argue this is fundamentally deceptive. While high-frequency tokens succeed in generation tasks due to implicit information leakage, isolating pure semantic information at ultra-low frame rates strips away the finegrained articulation and micro-dynamics essential for ODE-based generation. Empirically validating this requires extreme compression without sacrificing WER -- a methodological bottleneck, as standard fixed-stride downsampling arbitrarily truncates phonetic boundaries. To overcome this, we develop a dynamic compression tokenizer that intelligently aligns representations with semantic boundaries, achieving ultra-low frame rates with exceptionally low WER. Using these isolated "pure" semantic tokens, we expose the WER trap: when conditioning generative models -- even with oracle duration alignments -- the reconstructed speech suffers from severe articulation blur and is rendered acoustically unintelligible. Our findings demonstrate that semantic categorization rewarded by low WER is inherently orthogonal to the continuous phonetic trajectories required for synthesis, shattering the illusion of the unified token and advocating for explicitly decoupled speech representations.

eess.AS↗
arXiv2026

Dual Prototype-Conditioned Diffusion Model for Scalable Multi-Class Unsupervised Anomaly Detection in Large Category Spaces

Multi-class anomaly detection aims to build unified models across diverse product categories. However, as the number of categories grows, its performance often degrades due to increasingly complex and heterogeneous normal distributions. To address this challenge, we propose DPDiff-AD, a Dual Prototype-conditioned Diffusion model for large-scale multi-class Anomaly Detection. DPDiff-AD models heterogeneous normal distributions through complementary local and global prototypes. Local prototypes capture representative fine-grained structural patterns via nearest-prototype aggregation, while global prototypes regulate holistic feature geometry through optimal transport regularization. Together, these dual-scale representations define a structured normality space. This space is refined through diffusion-based reconstruction conditioned on both local and global prototypes via prototype-aware attention. By jointly leveraging dual prototypes during generation, DPDiff-AD achieves precise normality modeling, preserves structured separability as category cardinality grows, and enables scalable anomaly discrimination. Extensive experiments across five benchmarks demonstrate the effectiveness and scalability of DPDiff-AD. On the 160-category large-scale dataset, it improves image- and pixel-level AUROC by 5.3 and 2.9 points over the previous state-of-the-art method Dinomaly+, while maintaining stable performance as category cardinality increases.

cs.CV↗
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

DuplexSLA: A Full-Duplex Spoken Language Model with Synchronized Speech, Language, and Action

Recent advances in spoken dialogue language models have shifted from turn-based to full-duplex designs, where the model continuously listens to the user while generating responses. However, existing duplex backbones still lack a native channel for in-conversation planning and tool calling, leaving real-time agentic behaviour either tied to turn boundaries or relegated to an external cascade. We propose DuplexSLA, a native full-duplex Speech-Language-Action foundation model that decodes assistant audio together with a structured action stream on a shared 160 ms chunk timeline. DuplexSLA is built on a dual-stream three-channel formulation: a continuous user audio channel, a discrete assistant audio channel, and a rate-limited textual action channel, all decoded jointly by a single backbone, so that listening, speaking, planning, and tool calling unfold on one shared clock. Two capabilities define the model: (1) semantic-driven turn-taking control, where interruption, pause, and backchannel are handled inside the same backbone instead of by an external semantic VAD; and (2) in-conversation planning and tool calling, where planning text and structured tool calls are emitted on the action channel without halting assistant audio, so that multi-action and backchannel-triggered tool use are interleaved with ongoing speech. To evaluate these capabilities together, we further construct DuplexSLA-Bench, a duplex benchmark covering pause, interrupt, and backchannel turn-taking together with three styles of in-conversation tool calling. Our project page, interactive demos, and the DuplexSLA-Bench evaluation suite are publicly available at https://github.com/hyzhang24/DuplexSLA.

eess.AS↗
arXiv2026

Step-Audio-R1.5 Technical Report

Recent advancements in large audio language models have extended Chain-of-Thought (CoT) reasoning into the auditory domain, enabling models to tackle increasingly complex acoustic and spoken tasks. To elicit and sustain these extended reasoning chains, the prevailing paradigm -- driven by the success of text-based reasoning models -- overwhelmingly relies on Reinforcement Learning with Verified Rewards (RLVR). However, as models are strictly optimized to distill rich, continuous auditory contexts into isolated, verifiable text labels, a fundamental question arises: are we fostering true audio intelligence, or merely reducing a continuous sensory medium into a discrete puzzle? We identify this as the "verifiable reward trap." While RLVR yields remarkable scores on standardized objective benchmarks, it systematically degrades the real-world conversational feel of audio models. By prioritizing isolated correctness over acoustic nuance, RLVR reduces dynamic interactions to mechanical "answering machines," severely compromising prosodic naturalness, emotional continuity, and user immersion, particularly in long-turn dialogues. To bridge the gap between mechanical objective verification and genuine sensory empathy, we introduce Step-Audio-R1.5, marking a paradigm shift toward Reinforcement Learning from Human Feedback (RLHF) in audio reasoning. Comprehensive evaluations demonstrate that Step-Audio-R1.5 not only maintains robust analytical reasoning but profoundly transforms the interactive experience, redefining the boundaries of deeply immersive long-turn spoken dialogue.

eess.AS↗
arXiv2026

Weak Harnack inequality and Cartan property for nonlocal $W^{s,1}$-minimizers

We establish a weak Harnack inequality for nonlocal $W^{s,1}$-subminimizers in a complete, connected, doubling metric measure space where $0<s<1$. As a corollary, we prove that $W^{s,1}$-subminimizers are semicontinuous, up to a suitable choice of pointwise representative. We then prove \emph{Cartan-type properties} for $W^{s,1}$-superminimizers. The theory turns out to be mostly analogous with the local case of BV super- and subminimizers. Our results seem to be new even in the classical Euclidean setting.

math.AP↗
arXiv2026

The Silent Thought: Modeling Internal Cognition in Full-Duplex Spoken Dialogue Models via Latent Reasoning

During conversational interactions, humans subconsciously engage in concurrent thinking while listening to a speaker. Although this internal cognitive processing may not always manifest as explicit linguistic structures, it is instrumental in formulating high-quality responses. Inspired by this cognitive phenomenon, we propose a novel Full-duplex LAtent and Internal Reasoning method named FLAIR that conducts latent thinking simultaneously with speech perception. Unlike conventional "thinking" mechanisms in NLP, which require post-hoc generation, our approach aligns seamlessly with spoken dialogue systems: during the user's speaking phase, it recursively feeds the latent embedding output from the previous step into the next step, enabling continuous reasoning that strictly adheres to causality without introducing additional latency. To enable this latent reasoning, we design an Evidence Lower Bound-based objective that supports efficient supervised finetuning via teacher forcing, circumventing the need for explicit reasoning annotations. Experiments demonstrate the effectiveness of this think-while-listening design, which achieves competitive results on a range of speech benchmarks. Furthermore, FLAIR robustly handles conversational dynamics and attains competitive performance on full-duplex interaction metrics.

eess.AS↗
arXiv2026

Virtual reality for large-scale laboratories based on colorized point clouds: design and pedagogical impact

Effective laboratory training is essential in engineering education, yet conventional on-site instruction is often constrained by time, accessibility, and safety considerations. To ad-dress these challenges, this study presents the design, implementation, and evaluation of a web-based virtual reality (WebVR) representation of a large-scale engineering laboratory constructed from massive colorized point cloud data. This study proposes a novel WebVR approach that integrates Unity and Potree for high-fidelity point-cloud visualization com-bined with advanced interactive capabilities in a browser-based virtual laboratory. It supports immersive first-person exploration, guided navigation, interactive hotspots con-veying equipment and safety information, as well as emergency evacuation simulations. The usability, usefulness, and acceptance of the virtual laboratory were evaluated through an anonymous questionnaire administered to students and laboratory staff. User evalua-tion results indicated consistently positive feedback, with 100% of respondents rating the interface/navigation and visual/interactive content as good or excellent, 88.6% identifying scene realism as the biggest system strength (the most frequently selected), 74.3% reporting significantly higher engagement compared with traditional online laboratory training, and 82.9% indicating they would definitely recommend the system as a learning resource. In addition, a thematic analysis of qualitative feedback was performed to inform future enhancements of the WebVR environment. Overall, the findings demonstrate that the WebVR-based virtual laboratory can effectively complement conventional on-site labora-tory instruction, offering a scalable, accessible, and low-risk platform that enhances learning experiences in engineering education.

cs.HC↗
arXiv2026

Do Ambient Backscatter Communication Receivers Require Low-Noise Amplifiers?

In ambient backscatter communication (AmBC), strong direct interference from the ambient source poses a major challenge to reliable symbol detection. Although previous studies have shown that employing a low-noise amplifier (LNA) in conventional point-to-point communication improves symbol detection performance at low-to-moderate transmission power, it remains unclear whether this improvement also holds for AmBC. To respond it, in this work, we investigate the symbol detection performance of an AmBC receiver that is equipped with an LNA and adopts the energy detection (ED) to recover tag's information. Particularly, we first propose a new AmBC symbol detection framework that incorporates LNA parameters. On this basis, we derive the bit error rate (BER) of the ED and employ the deflection coefficient (DC) to demonstrate that the detection performance can be enhanced by the LNA at low-to-moderate ambient source transmission power. Then, we derive the near-optimal detection threshold to minimize the BER and propose a method to estimate the required parameters for this threshold by leveraging the tag's pilot symbols.

cs.IT↗
arXiv2026

DepFlow: Disentangled Speech Generation to Mitigate Semantic Bias in Depression Detection

Speech is a scalable and non-invasive biomarker for early mental health screening. However, widely used depression datasets like DAIC-WOZ exhibit strong coupling between linguistic sentiment and diagnostic labels, encouraging models to learn semantic shortcuts. As a result, model robustness may be compromised in real-world scenarios, such as Camouflaged Depression, where individuals maintain socially positive or neutral language despite underlying depressive states. To mitigate this semantic bias, we propose DepFlow, a three-stage depression-conditioned text-to-speech framework. First, a Depression Acoustic Encoder learns speaker- and content-invariant depression embeddings through adversarial training, achieving effective disentanglement while preserving depression discriminability (ROC-AUC: 0.693). Second, a flow-matching TTS model with FiLM modulation injects these embeddings into synthesis, enabling control over depressive severity while preserving content and speaker identity. Third, a prototype-based severity mapping mechanism provides smooth and interpretable manipulation across the depression continuum. Using DepFlow, we construct a Camouflage Depression-oriented Augmentation (CDoA) dataset that pairs depressed acoustic patterns with positive/neutral content from a sentiment-stratified text bank, creating acoustic-semantic mismatches underrepresented in natural data. Evaluated across three depression detection architectures, CDoA improves macro-F1 by 9%, 12%, and 5%, respectively, consistently outperforming conventional augmentation strategies in depression Detection. Beyond enhancing robustness, DepFlow provides a controllable synthesis platform for conversational systems and simulation-based evaluation, where real clinical data remains limited by ethical and coverage constraints.

cs.CL↗