arXiv subjects

Jin Li

Publications and source records attributed to Jin Li.

At least 55 records · Page 3Linked to original sources

arXiv2026

Independent-Component-Based Encoding Models of Brain Activity During Story Comprehension

Encoding models provide a powerful framework for linking continuous stimulus features to neural activity; however, traditional voxelwise approaches are limited by measurement noise, inter-subject variability, and redundancy arising from spatially correlated voxels encoding overlapping neural signals. Here, we propose an independent component (IC)-based encoding framework that dissociates stimulus-driven and noise-driven signals in fMRI data. We decompose continuous fMRI data from naturalistic story listening into ICs using one subset of the data, and train encoding models on independent data to predict IC time series from large language model representations of linguistic input. Across subjects, a subset of ICs exhibited consistently high predictivity. These ICs were spatially and temporally consistent across subjects and included cognitive networks known to respond during story listening (auditory and language). Auditory component time series were strongly correlated with acoustic stimulus features, highlighting the interpretability of identified component time series. Components identified as noise or motion-related artifacts by ICA-AROMA showed uniformly poor predictive performance, confirming that highly predicted components reflect genuine stimulus-related neural signals rather than confounds. Overall, IC-based encoding models enable analyses at the level of functional networks, accommodating the variability in network locations across individuals and providing interpretable results that are easy to compare across subjects. Code provided at: https://github.com/kamyahari/IC-Encoding-Models.git

cs.CL↗

arXiv2026

Constraining Kerr supermassive black hole properties using gravitational waves from inspiraling stellar-mass binary black holes

We study the capability of future space-based gravitational-wave (GW) detectors to constrain supermassive black hole (SMBH) properties through observations of inspiraling stellar-mass binary black holes (BBHs) orbiting them. Focusing on stable hierarchical triple systems, we model the BBH motion in Kerr spacetime and compute the modulated GW signals using the post Newtonian waveform combined with moving-source transformation. Based on the LISA configuration and second-generation time delay interferometry technology, we estimate parameter uncertainties with the Fisher information matrix. Our results show that the outer semimajor axis has the strongest influence on parameter precision, while the SMBH spin and eccentricity mainly affect their own uncertainties. For high-SNR signals, the SMBH mass and orbital parameters can be measured with relative uncertainties on the order of $10^{-5}$, while the spin magnitude and its orientation can be constrained to within a few percentages. Applying the method to an M87*-like system, GW observations provide more precise measurements of the SMBH mass and spin compared with current electromagnetic observations, highlighting the potential of space-based GW astronomy to probe SMBH properties with high accuracy.

gr-qc↗

arXiv2026

Identifiability of $g$ mode Resonances in Eccentric Binary Neutron Stars with Multidetector Observations

$g$ mode resonances in eccentric binary neutron star systems are potential probes of internal stratification, superfluidity, composition gradients, and the equation of state. Although such weak dynamical tidal signatures are unlikely to be resolved with current detector sensitivities, third generation observations may make them accessible, in which case identifying the weak resonant phase shift would provide information beyond the bulk adiabatic tidal deformability. We build a four class dataset in an eccentric harmonic framework, containing point particle, adiabatic tide, resonant $g$ mode, and pure noise samples, and use Einstein Telescope (ET) and Cosmic Explorer (CE) detector data to test whether this weak resonant phase signature can be identified from noisy time domain strain. The ET, CE, and ET+CE deep learning models reach accuracies of $0.655$, $0.815$, and $0.897$, respectively. On the same simulated samples, the matched filtering method reaches lower accuracies of $0.514$, $0.677$, and $0.689$. This result arises from the fact that the resonant correction manifests as a weak phase morphology difference superimposed on the adiabatic tidal background, whereas matched filtering is sensitive only to the overall similarity. Hence, in the presence of weak phase differences, the neural classifier employed in deep learning is better able to learn these local phase and morphology features from the complete time domain strain segment. The results indicate that joint third generation observations improve the identifiability of weak internal mode phase information.

astro-ph.HE↗

arXiv2026

KFC-KWS: Keyframe Fusion with CTC for User-Defined Keyword Spotting

User-defined keyword spotting (KWS) enables personalized voice interaction by detecting user-specified keywords. A key challenge in this task is distinguishing target keywords from phonetically confusable alternatives. To address this challenge, we propose KFC-KWS, a multimodal framework that leverages connectionist temporal classification (CTC)-guided keyframe selection. Specifically, we exploit the peaky posterior distributions of CTC to identify high-confidence phoneme frames, enabling precise alignment across audio, phoneme, and text modalities. These keyframes are then fused with full-utterance representations through cross-attention to capture both local discriminative cues and global contextual information. On LibriPhrase, KFC-KWS achieves the best-balanced performance (98.73% AUC) and substantially outperforms advanced baselines on the challenging hard subset (97.65% AUC and 7.75% EER), demonstrating its effectiveness in discriminating between highly confusable keywords.

cs.SD↗

arXiv2026

Escaping the KL Agreement Trap in On-Policy Distillation

On-policy distillation (OPD) provides dense token-level supervision by asking a teacher to score student-generated rollouts. However, when the student drifts into an unrecoverable prefix, the teacher may locally agree with the degraded state, producing low reverse KL but little corrective training signal. We identify this persistent regime as a low-KL agreement trap. Further analyses show that tokens during and after such traps produce less useful supervision signals. We propose KAT (KL Agreement Trap Termination), an online OPD termination rule that detects persistent low-KL agreement with a dynamic training-adaptive threshold. By filtering weak supervision from degenerate agreement, KAT improves avg@k accuracy by 2.66% and pass@k by 3.43% across four mathematical benchmarks, while reducing average rollout length by 59.73%.

cs.LG↗

arXiv2026

Detection of Multiband Lensed Gravitational Waves from Dark Matter Halos with Deep Learning

Lensed gravitational waves acquire amplitude and phase modulations when propagating through the gravitational potential of dark matter halos, producing interference structures in the observed waveform. However, these features are often difficult to identify in detector noise. In this work, we develop a deep learning framework for the automatic classification of lensed gravitational wave signals under multiband observations. We simulate binary neutron star signals observed by the space based detector DECIGO and the ground based Einstein Telescope, and construct five classes of data including pure noise, unlensed signals, and three lensed cases generated by the SIS, CIS, and NFW dark matter halo models. By comparing single detector and joint detector configurations, we evaluate the classification performance under different observational settings. The results show that multiband observations significantly improve the identification of lensed signals and reduce confusion among different lens models. This approach provides an efficient method for automated recognition of lensed gravitational waves in future multiband gravitational wave observations.

astro-ph.IM↗

arXiv2026

Unraveling the Hidden Dynamical Structure in Recurrent Neural Policies

Recurrent neural policies are widely used in partially observable control and meta-RL tasks. Their abilities to maintain internal memory and adapt quickly to unseen scenarios have offered them unparalleled performance when compared to non-recurrent counterparts. However, until today, the underlying mechanisms for their superior generalization and robustness performance remain poorly understood. In this study, by analyzing the hidden state domain of recurrent policies learned over a diverse set of training methods, model architectures, and tasks, we find that stable cyclic structures consistently emerge during interaction with the environment. Such cyclic structures share a remarkable similarity with \textit{limit cycles} in dynamical system analysis, if we consider the policy and the environment as a joint hybrid dynamical system. Moreover, we uncover that the geometry of such limit cycles also has a structured correspondence with the policies' behaviors. These findings offer new perspectives to explain many nice properties of recurrent policies: the emergence of limit cycles stabilizes both the policies' internal memory and the task-relevant environmental states, while suppressing nuisance variability arising from environmental uncertainty; the geometry of limit cycles also encodes relational structures of behaviors, facilitating easier skill adaptation when facing non-stationary environments.

cs.LG↗

arXiv2026

High-Throughput Discovery of Semimetallic Borophenes with Diverse Dirac States Via Transferable Tight-Binding Approach

Borophene has attracted extensive interest due to its structural flexibility and emergent topological electronic states. However, semimetallic borophenes hosting robust Dirac states remain rare among the large number of predicted allotropes. Here, we develop a transferable tight-binding framework for planar borophenes and combine it with a graph- and group-theory-based random generation strategy to perform high-throughput screening of 522 borophene candidates. Eight previously unreported semimetallic borophenes are identified, hosting diverse topological band crossings, including type-I and type-III Dirac cones, Dirac nodal lines, and quadratic nodal points. Notably, quadratic nodal-point semimetals are predicted in borophene for the first time. Symmetry analysis reveals crystalline-symmetry-protected Dirac states, while first-principles calculations confirm their dynamical and thermal stability. These findings establish borophene as a versatile platform for engineering emergent Dirac physics in two dimensions.

cond-mat.mtrl-sci↗

arXiv2026

Radiative MHD Studies of Solar Spicules: Driving Mechanisms and the Role of Background Magnetic Field

We conduct 2.5D radiative magnetohydrodynamic (MHD) simulations to investigate the driving mechanisms of the solar spicules in coronal holes and how the different background magnetic fields affect their formation. The simulation model includes the upper convection zone, photosphere, chromosphere, and lower corona. We run several cases with different background magnetic fields to primarily explore the effects of magnetic field strength and inclination on the characteristics of the formed spicules, such as the maximum height, lifetime, maximum upward velocity, and deceleration. According to the results, we find that a weaker background magnetic field tends to cause solar spicules that exhibit higher heights, longer lifetimes and faster maximum upward velocities. Additionally, when the background magnetic field inclines, the generated spicules incline accordingly; compared with those in a vertical magnetic field, the spicules in the inclined magnetic field are relatively longer, with smaller decelerations, longer lifetimes and lower maximum upward velocities. Meanwhile, by tracking and analyzing the formation processes of the spicules in two cases with different magnetic field strengths, we find that most of the spicules are mainly driven by shock waves induced by convective and turbulent motions around the solar surface, while less than one third of the spicules are primarily driven by high-velocity reconnection outflows. These results provide a more in-depth basis for the theoretical understanding of the driving mechanisms and formation processes of solar spicules.

astro-ph.SR↗

arXiv2026

AONA: A Comprehensive Architecture and Workflow Design for Global Agentic Collaboration

The rapid advancement of Large Language Models (LLMs) has established autonomous agents as the core vehicles for artificial intelligence applications. However, existing Internet infrastructures, primarily relying on TCP/IP and DNS, are designed for human-centric, host-to-host data transmission, inherently lacking the semantic awareness, dynamic capability discovery, and decentralized trust mechanisms required for autonomous agent interactions. To address these limitations and break the closed ecosystems of single vendors, this paper proposes AONA (Agentic Overlay Network Architecture), a novel overlay network architecture for the Internet of Agents (IoA). We first provide a multi-disciplinary scientific defense for multi-agent collaboration, demonstrating its theoretical necessity over single super-intelligence through the lenses of organizational economics, scaling principles, and the Price of Anarchy. AONA is then structured as a four-layer logical blueprint comprising the Base, Interconnection, Collaboration, and Application layers, which facilitates cross-protocol and cross-platform interoperability without disrupting the underlying physical network. To physically instantiate this blueprint, we design a distributed node infrastructure anchored by Management Root Nodes, Registry Service Nodes, Discovery Service Nodes, and Enterprise Intelligent Service Hubs for private domain integration. Finally, we detail the dynamic operational workflows-including zero-trust identity issuance, globally coordinated semantic taxonomy synchronization, intent-driven semantic discovery, and trusted metering for commercial settlement-that drive the network. This comprehensive architecture provides a robust, scalable, and secure foundation for the future of global agentic collaboration.

cs.NI↗

arXiv2026

Why Attend to Everything? Focus is the Key

Standard attention scales quadratically with sequence length. Efficient attention methods reduce this O(n^2) cost, but when retrofitted into pretrained models, they often degrade perplexity, downstream accuracy, or both. We introduce Focus, a method that learns which token pairs matter. Focus adds a small set of learnable centroids--as few as 148K parameters per layer--that act as gates: only token pairs belonging to the same centroid group attend to each other over long ranges. Focus is composable: it can be added to any pretrained model by training only the centroids while keeping all original weights frozen. Experiments show that composing Focus onto pretrained models yields zero degradation on downstream benchmarks across model sizes from 124M to 70B parameters and five attention architectures. Surprisingly, sparse Focus attention outperforms full attention at 124M scale (30.3 vs. 31.4 perplexity) and matches full attention when trained from scratch at 7B scale (13.82 vs. 13.89). Focus is also fast: top-k group membership gives a 2x speedup with better quality than the original pretrained model. Using our FlashAttention decomposition, Focus achieves an 8.6x speedup at 1M tokens without custom kernels.

cs.CL↗

arXiv2026

Probe-then-Plan: Environment-Aware Planning for Industrial E-commerce Search

Modern e-commerce search is evolving to resolve complex user intents. While Large Language Models (LLMs) offer strong reasoning, existing LLM-based paradigms face a fundamental blindness-latency dilemma: query rewriting is agnostic to retrieval capabilities and real-time inventory, yielding invalid plans; conversely, deep search agents rely on iterative tool calls and reflection, incurring seconds of latency incompatible with industrial sub-second budgets. To resolve this conflict, we propose Environment-Aware Search Planning (EASP), reformulating search planning as a dynamic reasoning process grounded in environmental reality. EASP introduces a Probe-then-Plan mechanism: a lightweight Retrieval Probe exposes the retrieval snapshot, enabling the Planner to diagnose execution gaps and generate grounded search plans. The methodology comprises three stages: (1) Offline Data Synthesis: A Teacher Agent synthesizes diverse, execution-validated plans by diagnosing the probed environment. (2) Planner Training and Alignment: The Planner is initialized via Supervised Fine-Tuning (SFT) to internalize diagnostic capabilities, then aligned with business outcomes (conversion rate) via Reinforcement Learning (RL). (3) Adaptive Online Serving: A complexity-aware routing mechanism selectively activates planning for complex queries, ensuring optimal resource allocation. Extensive offline evaluations and online A/B testing on JD.com demonstrate that EASP significantly improves relevant recall and achieves substantial lifts in UCVR and GMV. EASP has been successfully deployed in JD.com's AI-Search system.

cs.AI↗

arXiv2026

MoZoo:Unleashing Video Diffusion power in animal fur and muscle simulation

The creation of cinematic-quality animal effects necessitates the precise modeling of muscle and fur dynamics, a process that remains both labor-intensive and computationally expensive within traditional production workflows. While generative diffusion models have shown promise in diverse artistic workflows, their capacity for high-fidelity animal simulation remains largely unexploited. We present MoZoo, a generative dynamics solver that bypasses conventional refinement to synthesize high-fidelity animal videos from coarse meshes under multimodal guidance. We propose Role-Aware RoPE (RAR-RoPE) which employs role-based index remapping to synchronize motion alignment while decoupling reference information via fixed temporal offsets. Complementing this, Asymmetric Decoupled Attention partitions the latent sequence to enforce a unidirectional information flow, effectively preventing feature interference and improving computational efficiency. To address the scarcity of high-quality training data, we introduce MoZoo-Data, a synthetic-to-real pipeline that leverages a rendering engine and an inverse mapping approach to construct a large-scale dataset of paired sequences. Furthermore, we establish MoZooBench, a comprehensive benchmark with 120 mesh-video pairs. Experimental results demonstrate that MoZoo achieves high-fidelity fur simulation across diverse animal skeletons and layouts, preserving superior temporal and structural consistency.

cs.GR↗

arXiv2026

POTSA: A Cross-Lingual Speech Alignment Framework for Speech-to-Text Translation

Speech Large Language Models have achieved breakthroughs in multilingual speech-to-text translation. However, existing approaches often overlook semantic commonalities across source languages, leading to biased translation performance. In this work, we propose POTSA (Parallel Optimal Transport for Speech Alignment), a new framework based on cross-lingual parallel speech pairs and Optimal Transport, designed to bridge high- and low-resource translation gaps. First, we introduce a Bias Compensation module to coarsely align initial speech representations. Second, we impose token-level OT constraints on a Q-Former using parallel pairs to establish fine-grained representation consistency. Then, we apply a layer scheduling strategy to focus OT constraints on semantically beneficial layers. Experiments on FLEURS show our method achieves SOTA performance, with +1.29 BLEU over five common languages and +2.93 BLEU on zero-shot languages, using only 10 hours of parallel speech per language.

cs.CL↗

arXiv2026

Constraints and detection capabilities of GW polarizations with space-based detectors in different TDI combinations

TDI is essential in space-based GW detectors, effectively reducing laser noise and improving detection precision. Space-based GW detectors provide a unique opportunity to probe GW polarizations, including possible additional modes that may signal deviations from general relativity and alternative gravity theories. In this study, we examine the impacts of second-generation TDI combinations on GW polarization detection by simulating LISA, Taiji, and TianQin, including realistic orbital effects such as link length and angle variations. Detector performance is assessed using sensitivity and power-law integrated curves, as well as the SNR of BBHs and phase transitions (PTs). For massive BBHs, the A and E channels typically offer the best sensitivity, while the X channel in TianQin is most effective for detecting additional polarizations. For stellar-mass BBHs, the $α$ channel provides the highest SNR for vector modes in LISA and Taiji specifically for lower-mass systems, while the A and E channels are optimal for higher masses or other polarizations. For PT signals, the X channel generally delivers the optimal performance, except in the low-peak-frequency regime below 1 mHz, where the U channel in TianQin becomes more sensitive. When considering additional polarizations, the X channel emerges as the most robust choice for TianQin, in contrast to LISA and Taiji, where the A and E channels provide strong capabilities for GW polarization tests. This distinction between LISA, Taiji, and TianQin represents a key result of the present work and has not been explicitly emphasized in previous studies. Our findings emphasize the importance of selecting high-sensitivity TDI combinations to enhance detection capabilities across different polarizations, deepening our insight into GW sources and the fundamental nature of spacetime.

gr-qc↗

arXiv2026

From Evidence-Based Medicine to Knowledge Graph: Retrieval-Augmented Generation for Sports Rehabilitation and a Domain Benchmark

Current medical retrieval-augmented generation (RAG) approaches overlook evidence-based medicine (EBM) principles, leading to two key gaps: (1) the lack of PICO alignment between queries and retrieved evidence, and (2) the absence of evidence hierarchy considerations during reranking. We present SR-RAG, an EBM-adapted GraphRAG framework that integrates the PICO framework into knowledge graph construction and retrieval, and proposes Bayesian Evidence Tier Reranking (BETR) to calibrate ranking scores by evidence grade without predefined weights. Validated in sports rehabilitation, we release a knowledge graph (357,844 nodes, 371,226 edges) and a benchmark of 1,637 QA pairs. SR-RAG achieves 0.812 evidence recall@10, 0.830 nugget coverage, 0.819 answer faithfulness, 0.882 semantic similarity, and 0.788 PICOT match accuracy, substantially outperforming five baselines. Five expert clinicians rated the system 4.66--4.84 on a 5-point Likert scale, and system rankings are preserved on a human-verified gold subset (n=80).

cs.CL↗

arXiv2026

Speed by Simplicity: A Single-Stream Architecture for Fast Audio-Video Generative Foundation Model

We present daVinci-MagiHuman, an open-source audio-video generative foundation model for human-centric generation. daVinci-MagiHuman jointly generates synchronized video and audio using a single-stream Transformer that processes text, video, and audio within a unified token sequence via self-attention only. This single-stream design avoids the complexity of multi-stream or cross-attention architectures while remaining easy to optimize with standard training and inference infrastructure. The model is particularly strong in human-centric scenarios, producing expressive facial performance, natural speech-expression coordination, realistic body motion, and precise audio-video synchronization. It supports multilingual spoken generation across Chinese (Mandarin and Cantonese), English, Japanese, Korean, German, and French. For efficient inference, we combine the single-stream backbone with model distillation, latent-space super-resolution, and a Turbo VAE decoder, enabling generation of a 5-second 256p video in 2 seconds on a single H100 GPU. In automatic evaluation, daVinci-MagiHuman achieves the highest visual quality and text alignment among leading open models, along with the lowest word error rate (14.60%) for speech intelligibility. In pairwise human evaluation, it achieves win rates of 80.0% against Ovi 1.1 and 60.9% against LTX 2.3 over 2000 comparisons. We open-source the complete model stack, including the base model, the distilled model, the super-resolution model, and the inference codebase.

cs.CV↗

arXiv2026

GenFacet: End-to-End Generative Faceted Search via Multi-Task Preference Alignment in E-Commerce

Faceted search acts as a critical bridge for navigating massive ecommerce catalogs, yet traditional systems rely on static rule-based extraction or statistical ranking, struggling with emerging vocabulary, semantic gaps, and a disconnect between facet selection and underlying retrieval. In this paper, we introduce GenFacet, an industrial-grade, end-to-end generative framework deployed at JD.com. GenFacet reframes faceted search as two coupled generative tasks within a unified Large Language Model: Context-Aware Facet Generation, which dynamically synthesizes trend-responsive navigation options, and Intent-Driven Query Rewriting, which translates user interactions into precise search queries to close the retrieval loop. To bridge the gap between generative capabilities and search utility, we propose a novel multi-task training pipeline combining teacher-student distillation with GRPO. This aligns the model with complex user preferences by directly optimizing for downstream search satisfaction. Validated on China's largest selfoperated e-commerce platform via rigorous offline evaluations and online A/B tests, GenFacet demonstrated substantial improvements. Specifically, online results reveal a relative increase of 42.0% in facet Click-Through Rate (CTR) and 2.0% in User Conversion Rate (UCVR). These outcomes provide strong evidence of the benefits of generative methods for improving query understanding and user engagement in large-scale information retrieval systems.

cs.IR↗