arXiv subjects

Jian Liu

Publications and source records attributed to Jian Liu.

At least 145 records · Page 8Linked to original sources

arXiv2025

StreamingAssistant: Efficient Visual Token Pruning for Accelerating Online Video Understanding

Online video understanding is essential for applications like public surveillance and AI glasses. However, applying Multimodal Large Language Models (MLLMs) to this domain is challenging due to the large number of video frames, resulting in high GPU memory usage and computational latency. To address these challenges, we propose token pruning as a means to reduce context length while retaining critical information. Specifically, we introduce a novel redundancy metric, Maximum Similarity to Spatially Adjacent Video Tokens (MSSAVT), which accounts for both token similarity and spatial position. To mitigate the bidirectional dependency between pruning and redundancy, we further design a masked pruning strategy that ensures only mutually unadjacent tokens are pruned. We also integrate an existing temporal redundancy-based pruning method to eliminate temporal redundancy of the video modality. Experimental results on multiple online and offline video understanding benchmarks demonstrate that our method significantly improves the accuracy (i.e., by 4\% at most) while incurring a negligible pruning latency (i.e., less than 1ms). Our full implementation will be made publicly available.

cs.CV↗

arXiv2025

Self-Error Correcting Method for Magnetic-Array-Type Current Sensors in Multi-Core Cable Applications

Data-driven methods enable online assessment of error states in magnetic-array-type current sensors, and long-term measurement stability can be enhanced through further self-error correction. However, when the magnetic-array-type current sensors are applied to multi-conductor systems such as multi-core cables, the time-varying correlations among conductor currents may degrade the performance of multi-latent-variable data-driven models for error evaluation. To address this issue, this paper proposes a robust self-error correcting method for magnetic-array-type current sensors even under significant variations in phase current correlations (e.g., large fluctuations in three-phase current imbalance). By incorporating phase current decoupling and principal component analysis (PCA), the correlation analysis of multi-latent variables (i.e., multi-conductor currents) is transformed into a single-latent-variable (corresponding to single phase current) modeling problem. Experimental results demonstrate that the proposed method effectively detects error drifts of magnetic field sensors as low as $2\times10^{-3}$ in relative error and $2\times10^{-3}$ rad in phase error. Accurate evaluation and correction of each magnetic field sensor's error drifts substantially eliminates the overall error drift in the magnetic-array-type current sensor, validating the feasibility and effectiveness of the proposed self-error correcting method.

physics.ins-det↗

arXiv2025

Probing ice-rule-breaking transition in $\rm{Dy_2Ti_2O_7}$ thin film by proximitized transport and magnetic torque

While the spin ice state of bulk pyrochlores such as $\rm{Dy_2Ti_2O_7}$ and $\rm{Ho_2Ti_2O_7}$ has been extensively studied in the last several decades due to its unique degenerate ground state and emergent monopole excitation, whether it survives in the thin-film form remains a mystery. The limited volume of thin-film sample makes it challenging to study the intrinsic magnetic properties. Here, we synthesized 18nm-thick $\rm{Dy_2Ti_2O_7}$ thin film on YSZ (Yttria-stabilized Zirconia with 9.5 mol% $\rm{Y_2O_3}$) substrate and capped it by a thin conductive $\rm{Bi_2Ir_2O_7}$ layer, and performed the proximitized magnetoresistance measurements. Our study found that the ice-rule-breaking phase transition survives but with a modified effective nearest-neighbor interaction ($\rm{J_{eff}}=$ 1.054 K) and distorted Ising spin axes ($\rmε=+0.051)$ compared to the bulk crystal. The results are supported by the simultaneously measured capacitive torque magnetometry. Our study demonstrates that proximitized transport is an effective tool for thin films of insulating frustrated magnets.

cond-mat.mtrl-sci↗

arXiv2025

The Power of Many: Synergistic Unification of Diverse Augmentations for Efficient Adversarial Robustness

Adversarial perturbations pose a significant threat to deep learning models. Adversarial Training (AT), the predominant defense method, faces challenges of high computational costs and a degradation in standard performance. While data augmentation offers an alternative path, existing techniques either yield limited robustness gains or incur substantial training overhead. Therefore, developing a defense mechanism that is both highly efficient and strongly robust is of paramount importance.In this work, we first conduct a systematic analysis of existing augmentation techniques, revealing that the synergy among diverse strategies -- rather than any single method -- is crucial for enhancing robustness. Based on this insight, we propose the Universal Adversarial Augmenter (UAA) framework, which is characterized by its plug-and-play nature and training efficiency. UAA decouples the expensive perturbation generation process from model training by pre-computing a universal transformation offline, which is then used to efficiently generate unique adversarial perturbations for each sample during training.Extensive experiments conducted on multiple benchmarks validate the effectiveness of UAA. The results demonstrate that UAA establishes a new state-of-the-art (SOTA) for data-augmentation-based adversarial defense strategies , without requiring the online generation of adversarial examples during training. This framework provides a practical and efficient pathway for building robust models,Our code is available in the supplementary materials.

cs.CV↗

arXiv2025

Multi-scale symmetry analysis in molecular structures

Topological data analysis (TDA), as a relatively recent approach, has demonstrated great potential in capturing the intrinsic and robust structural features of complex data. While persistent homology, as a core tool of TDA, focuses on characterizing geometric shapes and topological structures, the automorphism groups of Vietoris-Rips complexes can capture the structured symmetry features of data. In this work, we propose a multi-scale symmetry analysis approach that leverages persistent automorphism modules to quantify variations in symmetries across scales. By modifying the category of graphs and constructing a suitable functor from the graph category to the category of modules, we ensure that the persistent automorphism module forms a genuine persistence module. Furthermore, we apply this framework to the structural analysis of fullerenes, predicting the stability of 12 fullerene molecules with a competitive correlation coefficient of 0.979.

math.CO↗

arXiv2025

BLM$_1$: A Boundless Large Model for Cross-Space, Cross-Task, and Cross-Embodiment Learning

Multimodal large language models (MLLMs) have advanced vision-language reasoning and are increasingly deployed in embodied agents. However, significant limitations remain: MLLMs generalize poorly across digital-physical spaces and embodiments; vision-language-action models (VLAs) produce low-level actions yet lack robust high-level embodied reasoning; and most embodied large language models (ELLMs) are constrained to digital-space with poor generalization to the physical world. Thus, unified models that operate seamlessly across digital and physical spaces while generalizing across embodiments and tasks remain absent. We introduce the \textbf{Boundless Large Model (BLM$_1$)}, a multimodal spatial foundation model that preserves instruction following and reasoning, incorporates embodied knowledge, and supports robust cross-embodiment control. BLM$_1$ integrates three key capabilities -- \textit{cross-space transfer, cross-task learning, and cross-embodiment generalization} -- via a two-stage training paradigm. Stage I injects embodied knowledge into the MLLM through curated digital corpora while maintaining language competence. Stage II trains a policy module through an intent-bridging interface that extracts high-level semantics from the MLLM to guide control, without fine-tuning the MLLM backbone. This process is supported by a self-collected cross-embodiment demonstration suite spanning four robot embodiments and six progressively challenging tasks. Evaluations across digital and physical benchmarks show that a single BLM$_1$ instance outperforms four model families -- MLLMs, ELLMs, VLAs, and GMLMs -- achieving $\sim\!\textbf{6%}$ gains in digital tasks and $\sim\!\textbf{3%}$ in physical tasks.

cs.AI↗

arXiv2025

Ground-state properties of finite nuclei in relativistic Hartree-Bogoliubov theory with an improved quark mass density-dependent model

A relativistic Hartree-Bogoliubov (RHB) model based on quark-meson coupling is developed, with a new parametrization derived from experimental observables. Using this model, we systematically investigate the ground-state properties of even-even nuclei spanning $8\leq Z\leq118$, including binding energies, quadrupole deformations, root-mean-square (rms) charge radii, two-nucleon separation energies, two-nucleon shell gaps, and $α$-decay energies. Comparisons with available experimental data demonstrate that this subnucleon-based RHB model reliably describes the ground-state properties of finite nuclei.

nucl-th↗

arXiv2025

Every Step Evolves: Scaling Reinforcement Learning for Trillion-Scale Thinking Model

We present Ring-1T, the first open-source, state-of-the-art thinking model with a trillion-scale parameter. It features 1 trillion total parameters and activates approximately 50 billion per token. Training such models at a trillion-parameter scale introduces unprecedented challenges, including train-inference misalignment, inefficiencies in rollout processing, and bottlenecks in the RL system. To address these, we pioneer three interconnected innovations: (1) IcePop stabilizes RL training via token-level discrepancy masking and clipping, resolving instability from training-inference mismatches; (2) C3PO++ improves resource utilization for long rollouts under a token budget by dynamically partitioning them, thereby obtaining high time efficiency; and (3) ASystem, a high-performance RL framework designed to overcome the systemic bottlenecks that impede trillion-parameter model training. Ring-1T delivers breakthrough results across critical benchmarks: 93.4 on AIME-2025, 86.72 on HMMT-2025, 2088 on CodeForces, and 55.94 on ARC-AGI-1. Notably, it attains a silver medal-level result on the IMO-2025, underscoring its exceptional reasoning capabilities. By releasing the complete 1T parameter MoE model to the community, we provide the research community with direct access to cutting-edge reasoning capabilities. This contribution marks a significant milestone in democratizing large-scale reasoning intelligence and establishes a new baseline for open-source model performance.

cs.CL↗

arXiv2025

Mesh-RFT: Enhancing Mesh Generation via Fine-grained Reinforcement Fine-Tuning

Existing pretrained models for 3D mesh generation often suffer from data biases and produce low-quality results, while global reinforcement learning (RL) methods rely on object-level rewards that struggle to capture local structure details. To address these challenges, we present Mesh-RFT, a novel fine-grained reinforcement fine-tuning framework that employs Masked Direct Preference Optimization (M-DPO) to enable localized refinement via quality-aware face masking. To facilitate efficient quality evaluation, we introduce an objective topology-aware scoring system to evaluate geometric integrity and topological regularity at both object and face levels through two metrics: Boundary Edge Ratio (BER) and Topology Score (TS). By integrating these metrics into a fine-grained RL strategy, Mesh-RFT becomes the first method to optimize mesh quality at the granularity of individual faces, resolving localized errors while preserving global coherence. Experiment results show that our M-DPO approach reduces Hausdorff Distance (HD) by 24.6% and improves Topology Score (TS) by 3.8% over pre-trained models, while outperforming global DPO methods with a 17.4% HD reduction and 4.9% TS gain. These results demonstrate Mesh-RFT's ability to improve geometric integrity and topological regularity, achieving new state-of-the-art performance in production-ready mesh generation. Project Page: https://hitcslj.github.io/mesh-rft/.

cs.CV↗

arXiv2025

FerretNet: Efficient Synthetic Image Detection via Local Pixel Dependencies

The increasing realism of synthetic images generated by advanced models such as VAEs, GANs, and LDMs poses significant challenges for synthetic image detection. To address this issue, we explore two artifact types introduced during the generation process: (1) latent distribution deviations and (2) decoding-induced smoothing effects, which manifest as inconsistencies in local textures, edges, and color transitions. Leveraging local pixel dependencies (LPD) properties rooted in Markov Random Fields, we reconstruct synthetic images using neighboring pixel information to expose disruptions in texture continuity and edge coherence. Building upon LPD, we propose FerretNet, a lightweight neural network with only 1.1M parameters that delivers efficient and robust synthetic image detection. Extensive experiments demonstrate that FerretNet, trained exclusively on the 4-class ProGAN dataset, achieves an average accuracy of 97.1% on an open-world benchmark comprising 22 generative models. Our code and datasets are publicly available at https://github.com/xigua7105/FerretNet.

cs.CV↗

arXiv2025

Classifying thick subcategories over a Koszul complex via the curved BGG correspondence

In this work we classify the thick subcategories of the bounded derived category of dg modules over a Koszul complex on any list of elements in a regular ring. This simultaneously recovers a theorem of Stevenson when the list of elements is a regular sequence and the classification of thick subcategories for an exterior algebra over a field (via the BGG correspondence). One of the major ingredients is a classification of thick tensor submodules of perfect curved dg modules over a commutative noetherian graded ring concentrated in even degrees, recovering a theorem of Hopkins and Neeman. We give several consequences of the classification result over a Koszul complex, one being that the lattice of thick subcategories of the bounded derived category is fixed by Grothendieck duality.

math.AC↗

arXiv2025

FlexQuant: A Flexible and Efficient Dynamic Precision Switching Framework for LLM Quantization

The rapid advancement of large language models (LLMs) has exacerbated the memory bottleneck due to the widening gap between model parameter scaling and hardware capabilities. While post-training quantization techniques effectively reduce memory overhead, existing methods predominantly rely on static quantization strategies, which struggle to adapt to dynamic workloads. To address this, we propose FlexQuant, a dynamic precision-switching framework that optimizes the trade-off between inference speed and accuracy. Leveraging model perplexity entropy and Kullback-Leibler divergence, FlexQuant enables fine-grained, layer-wise mixed-precision quantization and dynamically adjusts bit-widths during each token generation. FlexQuant provides a comprehensive analysis of quantization strategies, introduces a precision requirement model for optimal switching, and implements efficient fine-grained precision management. Evaluations demonstrate that FlexQuant achieves a 1.3x end-to-end speedup across diverse language tasks with negligible accuracy loss introduced. This framework offers a flexible and adaptive solution for efficient LLM deployment. Code is released at https://github.com/ZongwuWang/FlexQuant.git.

cs.LG↗

arXiv2025

Topological Sequence Analysis of Genomes: Category theory Approaches

Sequence data, such as DNA, RNA, and protein sequences, exhibit intricate, multi-scale structures that pose significant challenges for conventional analysis methods, particularly those relying on alignment or purely statistical representations. In this work, we introduce category-based topological sequence analysis (CTSA ) of genomes. CTSA models a sequence as a resolution category, capturing its hierarchical structure through a categorical construction. Substructure complexes are then derived from this categorical representation, and their persistent homology is computed to extract multi-scale topological features. Our models depart from traditional alignment-free approaches by incorporating structured mathematical formalisms rooted in sequence topology. The resulting topological signatures provide informative representations across a variety of tasks, including the phylogenetic analysis of SARS-CoV-2 variants and the prediction of protein-nucleic acid binding affinities. Comparative studies were carried out against six state-of-the-art methods. Experimental results demonstrate that CTSA achieves excellent and consistent performance in these tasks, suggesting its general applicability and robustness. Beyond sequence analysis, the proposed framework opens new directions for the integration of categorical and homological theories for biological sequence analysis.

q-bio.GN↗

arXiv2025

Auto-Connect: Connectivity-Preserving RigFormer with Direct Preference Optimization

We introduce Auto-Connect, a novel approach for automatic rigging that explicitly preserves skeletal connectivity through a connectivity-preserving tokenization scheme. Unlike previous methods that predict bone positions represented as two joints or first predict points before determining connectivity, our method employs special tokens to define endpoints for each joint's children and for each hierarchical layer, effectively automating connectivity relationships. This approach significantly enhances topological accuracy by integrating connectivity information directly into the prediction framework. To further guarantee high-quality topology, we implement a topology-aware reward function that quantifies topological correctness, which is then utilized in a post-training phase through reward-guided Direct Preference Optimization. Additionally, we incorporate implicit geodesic features for latent top-k bone selection, which substantially improves skinning quality. By leveraging geodesic distance information within the model's latent space, our approach intelligently determines the most influential bones for each vertex, effectively mitigating common skinning artifacts. This combination of connectivity-preserving tokenization, reward-guided fine-tuning, and geodesic-aware bone selection enables our model to consistently generate more anatomically plausible skeletal structures with superior deformation properties.

cs.CV↗

arXiv2025

Generation of singularity categories and infinite injective dimension locus via annihilation of cohomologies

Let R be a commutative Noetherian ring. We establish a close relationship between the strong generation of the singularity category of R and the nonvanishing of the annihilator of the singularity category of R. As an application, we prove that the singularity category of R has a strong generator if and only if the annihilator of the singularity category of R is nonzero when R is a Noetherian domain with Krull dimension at most one. We introduce the notion of the co-cohomological annihilator of modules. If the category of finitely generated R-modules has a strong generator, we show that the infinite injective dimension locus of a finitely generated R-module M is closed, with the defining ideal given by the co-cohomological annihilator of M. Finally, we provide a connection between the existence of an extension generator of the category of finitely generated R-modules and the finiteness of the Krull dimension of R.

math.AC↗

arXiv2025

Ignition Phase : Standard Training for Fast Adversarial Robustness

Adversarial Training (AT) is a cornerstone defense, but many variants overlook foundational feature representations by primarily focusing on stronger attack generation. We introduce Adversarial Evolution Training (AET), a simple yet powerful framework that strategically prepends an Empirical Risk Minimization (ERM) phase to conventional AT. We hypothesize this initial ERM phase cultivates a favorable feature manifold, enabling more efficient and effective robustness acquisition. Empirically, AET achieves comparable or superior robustness more rapidly, improves clean accuracy, and cuts training costs by 8-25\%. Its effectiveness is shown across multiple datasets, architectures, and when augmenting established AT methods. Our findings underscore the impact of feature pre-conditioning via standard training for developing more efficient, principled robust defenses. Code is available in the supplementary material.

cs.LG↗

arXiv2025

Neural Brain: A Neuroscience-inspired Framework for Embodied Agents

The rapid evolution of artificial intelligence (AI) has shifted from static, data-driven models to dynamic systems capable of perceiving and interacting with real-world environments. Despite advancements in pattern recognition and symbolic reasoning, current AI systems, such as large language models, remain disembodied, unable to physically engage with the world. This limitation has driven the rise of embodied AI, where autonomous agents, such as humanoid robots, must navigate and manipulate unstructured environments with human-like adaptability. At the core of this challenge lies the concept of Neural Brain, a central intelligence system designed to drive embodied agents with human-like adaptability. A Neural Brain must seamlessly integrate multimodal sensing and perception with cognitive capabilities. Achieving this also requires an adaptive memory system and energy-efficient hardware-software co-design, enabling real-time action in dynamic environments. This paper introduces a unified framework for the Neural Brain of embodied agents, addressing two fundamental challenges: (1) defining the core components of Neural Brain and (2) bridging the gap between static AI models and the dynamic adaptability required for real-world deployment. To this end, we propose a biologically inspired architecture that integrates multimodal active sensing, perception-cognition-action function, neuroplasticity-based memory storage and updating, and neuromorphic hardware/software optimization. Furthermore, we also review the latest research on embodied agents across these four aspects and analyze the gap between current AI systems and human intelligence. By synthesizing insights from neuroscience, we outline a roadmap towards the development of generalizable, autonomous agents capable of human-level intelligence in real-world scenarios.

cs.RO↗

arXiv2025

MS-YOLO: Infrared Object Detection for Edge Deployment via MobileNetV4 and SlideLoss

Infrared imaging has emerged as a robust solution for urban object detection under low-light and adverse weather conditions, offering significant advantages over traditional visible-light cameras. However, challenges such as class imbalance, thermal noise, and computational constraints can significantly hinder model performance in practical settings. To address these issues, we evaluate multiple YOLO variants on the FLIR ADAS V2 dataset, ultimately selecting YOLOv8 as our baseline due to its balanced accuracy and efficiency. Building on this foundation, we present \texttt{MS-YOLO} (\textbf{M}obileNetv4 and \textbf{S}lideLoss based on YOLO), which replaces YOLOv8's CSPDarknet backbone with the more efficient MobileNetV4, reducing computational overhead by \textbf{1.5%} while sustaining high accuracy. In addition, we introduce \emph{SlideLoss}, a novel loss function that dynamically emphasizes under-represented and occluded samples, boosting precision without sacrificing recall. Experiments on the FLIR ADAS V2 benchmark show that \texttt{MS-YOLO} attains competitive mAP and superior precision while operating at only \textbf{6.7 GFLOPs}. These results demonstrate that \texttt{MS-YOLO} effectively addresses the dual challenge of maintaining high detection quality while minimizing computational costs, making it well-suited for real-time edge deployment in urban environments.

cs.CV↗