PyTorchKR Papers

최신 논문

총 523,801편
태그·기간 등 상세 필터

카테고리 탐색

arXiv:2510.00294v4cs.LG

Free Draft-and-Verification: Toward Lossless Parallel Decoding for Diffusion Large Language Models

Diffusion Large Language Models (dLLMs) have emerged as a compelling paradigm for language modeling, offering the unique potential for highly efficient inference via multi-token parallel decoding. However, unlocking this potential remains challenging. High-quality generation typically defaults to one-token-per-step static decoding, while existing parallel algorithms rely on heuristic confidence thresholds, introducing a fragile trade-off between decoding efficiency and generation quality. To overcome this, we introduce Free Draft-and-Verification (FreeDave), a novel training-free and model-free fast decoding algorithm. FreeDave capitalizes on the inherent properties of dLLMs by treating future token predictions as ``free drafts'' and rigorously self-verifying them in subsequent forward passes. Theoretically, FreeDave searches for a shorter decoding path that is guaranteed to reproduce the exact sequence generated by static decoding, achieving algorithmically lossless acceleration. With a sufficiently large draft budget and a verification frontier that does not retreat, it can use the fewest possible model forward passes among feasible paths. Extensive evaluations on math reasoning and code generation benchmarks demonstrate that FreeDave can boost the number of generated tokens per model forward (TPF) by up to 5.48$\times$ and tokens per second (TPS) by up to 3.94$\times$, while robustly maintaining generation quality and avoiding the task-specific performance degradation faced by heuristic methods. Code is available at https://github.com/cychomatica/FreeDave.

arXiv:2510.00279v4cs.LG

SLogic: Subgraph-Informed Logical Rule Learning for Knowledge Graph Completion

Logical rule-based methods offer an interpretable approach to knowledge graph completion (KGC) by capturing compositional relationships in the form of human-readable inference rules. While existing logical rule-based methods learn rule confidence scores, they typically assign a global weight to each rule schema, applied uniformly across the graph. This is a significant limitation, as a rule's importance often varies depending on the specific query instance. To address this, we introduce SLogic (Subgraph-Informed Logical Rule learning), a novel framework that assigns query-dependent scores to logical rules. The core of SLogic is a context-aware scoring function. This framework is in line with the specificity principle in commonsense reasoning. This function determines the importance of a rule by analyzing the subgraph locally defined by the query's head entity, thereby enabling a differentiated weighting of rules specific to their local query contexts. Extensive experiments on benchmark datasets show that SLogic achieves competitive performance among rule-based methods. It also generates query-dependent, human-readable logical rules that serve as explicit explanations for its inferences.

arXiv:2509.21961v3cs.RO

FlowDrive: moderated flow matching with data balancing for trajectory planning

Learning-based planners are sensitive to the long-tailed distribution of driving data. Common maneuvers dominate datasets, while dangerous or rare scenarios are sparse. This imbalance can bias models toward the frequent cases and degrade performance on critical scenarios. To tackle this problem, we compare balancing strategies for sampling training data and find reweighting by trajectory pattern an effective approach. We then present FlowDrive, a flow-matching trajectory planner that learns a conditional rectified flow to map noise directly to trajectory distributions with few flow-matching steps. We further introduce moderated, in-the-loop guidance that injects small perturbation between flow steps to systematically increase trajectory diversity while remaining scene-consistent. On nuPlan and the interaction-focused interPlan benchmarks, FlowDrive achieves state-of-the-art results among learning-based planners and approaches methods with rule-based refinements. After adding moderated guidance and light post-processing (FlowDrive*), it achieves overall state-of-the-art performance across nearly all benchmark splits. Our code is available at https://github.com/einsteinguang/flow_drive_planner.

arXiv:2509.21745v2eess.SY

Reinforcement Learning based Traffic Signal Control for Queue Minimization: From Lane-based to Laneless Traffic

Efficient traffic signal control (TSC) is essential for reducing congestion and delays. Traditional TSC methods struggle to capture the dynamics of laneless, heterogeneous traffic. We propose a reinforcement learning (RL) based adaptive TSC framework whose reward depends only on approach-level queue length, motivated by the difficulty of obtaining per-vehicle measurements in laneless traffic, and evaluate it using SUMO on a conventional lane-based, homogeneous (LBHO) intersection and on a non-orthogonal laneless, heterogeneous (LLHE) intersection. On the LBHO site, the lowest mean configuration has about 23\% lower mean queue length than a Dynamic Webster baseline, and the unselected Full-State configuration about 15\% lower. On the LLHE site, all RL configurations have 66-73\% lower mean queue length than a Dynamic Webster baseline under nominal demand, and 63--74\% lower under two higher-demand scenarios evaluated without retraining. Our algorithm also outperforms actuated control and a RESCO-DQN baseline on mean queue length. We additionally release a laneless, heterogeneous-vehicle SUMO framework integrated with RL based signal control, to our knowledge the first such open implementation on a current SUMO version.

arXiv:2509.19930v2cs.LG

How deep is your network? Deep vs. shallow learning of transfer operators

We introduce RaNNDy, a randomized neural network framework for learning transfer operators and their spectral decompositions. By fixing the weights of the hidden layers and training only the output layer, RaNNDy achieves significant reductions in computational costs -- often by two orders of magnitude -- with similar accuracy compared to state-of-the-art deep learning methods. The method is less sensitive to hyperparameter tuning and provides a closed-form solution for the output layer weights that directly represent eigenfunctions of the learned operator. Moreover, it is possible to estimate uncertainties associated with the computed spectral properties via ensemble learning. We demonstrate the efficacy of the proposed approach using diverse benchmark problems, including stochastic dynamical systems and protein folding processes, highlighting the strengths but also weaknesses of RaNNDy.

arXiv:2509.19401v2eess.SP

SpellerSSL: Self-Supervised Learning with P300 Aggregation for Speller BCIs

Electroencephalogram (EEG)-based P300 speller brain-computer interfaces (BCIs) face three main challenges: low signal-to-noise ratio (SNR), poor generalization, and time-consuming calibration. We propose SpellerSSL, a framework that combines self-supervised learning (SSL) with P300 aggregation to address these issues. First, we introduce an aggregation strategy to enhance SNR. Second, to achieve generalization in training, we employ a customized 1D U-Net backbone and pre-train the model on both cross-domain and in-domain EEG data. The pretrained model is subsequently fine-tuned with a lightweight ERP-Head classifier for P300 detection, which adapts the learned representations to subject-specific data. Our evaluations on calibration time demonstrate that combining the aggregation strategy with SSL significantly reduces the calibration burden per subject and improves robustness across subjects. Experimental results show that SSL learns effective EEG representations in both in-domain and cross-domain, with in-domain achieving a state-of-the-art character recognition rate of 94% with only 7 repetitions and the highest information transfer rate (ITR) of 21.86 bits/min on the public II-B dataset. Moreover, in-domain SSL with P300 aggregation reduces the required calibration size by 60% while maintaining a comparable character recognition rate. To the best of our knowledge, this is the first study to apply SSL to P300 spellers, highlighting its potential to improve both efficiency and generalization in speller BCIs and paving the way toward an EEG foundation model for P300 speller BCIs. The code is available at: https://github.com/Emotiv/SpellerSSL-NeurIPS2025

arXiv:2509.10080v3cs.CV

BEVTraj: Map-Free Multimodal Trajectory Prediction with Sensor-Derived BEV Context and Sparse Goal Reasoning

Trajectory prediction is a key task in autonomous driving, where a model must forecast plausible future motions of surrounding agents from dynamic histories and scene context. Many recent methods use high-definition (HD) maps to provide explicit road topology and semantic constraints, but such maps are costly to maintain, geographically limited, and difficult to update in changing environments. This paper presents Bird's-Eye View Trajectory Prediction (BEVTraj), a map-free trajectory prediction framework that directly uses BEV features constructed from raw multi-sensor data. Since sensor-derived BEV features are dense and only implicitly structured, BEVTraj employs deformable attention to selectively aggregate agent-relevant spatial context from sparse locations rather than processing the entire BEV grid uniformly. In addition, a Sparse Goal Candidate Proposal module combines overlapping mode-wise anchor clusters with attention-based goal prediction to generate compact, scene-adaptive goal hypotheses. These goals are decoded and iteratively refined into multimodal trajectories, while trajectory-aware mode scoring and temporal consistency modeling improve confidence estimation and suppress abrupt, physically implausible refinement updates. Experiments on nuScenes and Argoverse 2 Sensor show that BEVTraj achieves competitive results under a map-free setting without using pre-built HD maps, while ablation studies and qualitative analyses illustrate the roles of BEV context aggregation and sparse goal reasoning. The source code is available at https://github.com/Kongminsang/bevtraj.

arXiv:2509.07370v3cs.CL

PersonaFuse: A Situation-Aware Post-Training Framework for Adaptive LLM Behavior in Enterprise Tasks

Large language models (LLMs) are increasingly deployed in enterprise settings to support task execution and workflow automation. Despite the strong performance of general-purpose models on standard benchmarks, organizations often need to adapt them to domain-specific tasks and operational constraints in practice. However, existing post-training approaches introduce capability trade-offs and lack mechanisms for situational adaptation. In this work, we propose PersonaFuse, a post-training framework that enables situation-aware behavioral adaptation in LLMs. Grounded in Trait Activation Theory and operationalized through the Five-Factor Model, PersonaFuse dynamically adjusts model behavior based on task context using a mixture-of-experts architecture with trait-level experts and a context-aware routing mechanism. We evaluate PersonaFuse at three levels. At the capability level, it achieves balanced improvements across general reasoning, interaction quality, and safety without introducing trade-offs. At the workflow level, it improves task completion in enterprise scenarios involving tool use and structured execution across both B2B and B2C tasks. At the deployment level, we analyze model selection under cost and safety constraints using a capability-cost frontier framework, showing that models adapted with PersonaFuse achieve more favorable trade-offs and are consistently selected as optimal across deployment scenarios. PersonaFuse provides a practical framework for developing LLMs that are not only capable, but also adaptable and deployable in real-world enterprise environments.

arXiv:2509.07274v5cs.CL

LLM Analysis of 150+ Years of German Parliamentary Debates on Migration Reveals a Shift from Post-War Solidarity to Anti-Solidarity in the Last Decade

Migration has been a core topic in German political debate, from postwar expellee displacement to labor migration and recent refugee movements. Large-scale analysis of such political discourse has traditionally required extensive manual annotation, limiting coverage. Large language models (LLMs) offer a scalable alternative. Using a theory-driven annotation scheme, we examine how well LLMs annotate subtypes of solidarity and anti-solidarity in German parliamentary debates and whether the resulting labels support valid downstream inference. We first evaluate multiple LLMs across model size, prompting strategies, fine-tuning, historical versus contemporary data, and systematic errors. The strongest models, especially GPT-5 and gpt-oss-120B, achieve macro-F1 scores comparable to human agreement, although their systematic errors can bias downstream results. We therefore combine soft-label model outputs with Design-based Supervised Learning (DSL) to reduce bias in long-term trend estimates. Beyond the methodological evaluation, we interpret the resulting annotations from a social-scientific perspective across the Reichstag (1867-1933), West German Bundestag (1949-1990) and Re-Unified German Bundestag (1990-2025) corpus to trace trends in solidarity and anti-solidarity toward migrants, with detailed analysis of postwar Germany (1949-1957) and contemporary Germany (2009-2025). We find relatively high levels of solidarity in the postwar period, especially in group-based and compassionate forms, and a marked rise in anti-solidarity since 2015. We argue that LLMs can support large-scale social-scientific text analysis, but their outputs require rigorous validation and, where systematic errors are present, statistical correction.

arXiv:2508.21380v5cs.LG

The Algorithm Is Not the Behavior: Learned Priors Override Look-Ahead in a Chess-Playing Neural Network

Recent mechanistic work has uncovered learned algorithms within neural networks, from modular arithmetic to search and planning in game-playing agents. But does algorithmic structure guarantee algorithmic behavior? We investigate this in Leela Chess Zero, the strongest neural chess engine, where prior work identified learned look-ahead. By extending the logit lens to its move-selecting policy network, we discover that correct puzzle solutions--including immediate checkmates--often appear in intermediate layers but are systematically overridden in the final output, a phenomenon we term "forgotten puzzles". Replicating prior analyses on these positions, we find that look-ahead operates normally--future moves of the correct continuation are represented, causally important, and linearly decodable--ruling out a failure of the algorithm. Instead, late layers increasingly shift toward safe play over aggression. To test whether this shift drives the override, we steer the model against these preferences and recover 71.7% of forgotten puzzles, providing causal evidence that safety priors override algorithmically computed solutions. These findings show that algorithmic structure does not guarantee algorithmic behavior: a model can internally solve a problem and still output the wrong answer.

arXiv:2508.12969v2cs.CV

Compact Attention: Exploiting Structured Spatio-Temporal Sparsity for Fast Video Generation

The quadratic computational complexity of self-attention mechanisms imposes a critical constraint on transformer-based video generation, particularly for synthesizing ultra-long and high-definition sequences. While most sparse attention techniques employ fixed or heuristic patterns, they often struggle to accommodate the heterogeneous and layer-specific redundancies inherent in video diffusion transformers (VDiTs). We find that different attention heads consistently concentrate most of their attention mass in different coarse query-key regions, a behavior we call functional specialization. Although individual attention values vary with the generated content, these regions remain similar across prompts and random seeds. Leveraging these insights, we introduce Compact Attention, a hardware-aware sparse attention framework for efficient long-video synthesis. We quantify this stability and use it to construct searchable, reusable, head-specific masks. Based on the observation, we propose a deformable dual-window sparse pattern mechanism that flexibly approximates local and axial attention regions within one tile-aligned mask family. To avoid manually choosing a mask for each head, we develop a recall-driven offline auto-search algorithm that identifies high-quality sparse masks while leveraging temporal stability for mask reuse. Extensive evaluations on Wan2.1 (14B) and Hunyuan-Video demonstrate that Compact Attention achieves 2.25x/3.03x attention speedups over FlashAttention-2 and 1.20x/1.74x over FlashAttention-3, respectively, while maintaining visual quality and temporal consistency comparable to full-attention baselines. Project Page: https://yo-ava.github.io/Compact-Attention.github.io/

arXiv:2508.06765v2cs.LG

Efficient Collaborative LLM Fine-Tuning over Heterogeneous Mobile Devices via Many Backbones to One Side-Network Tuning

Collaboratively fine-tuning (FT) large language models (LLMs) over heterogeneous mobile devices fosters immense potential applications of personalized intelligence. However, such a vision faces critical system challenges. Existing federated LLM fine-tuning approaches remain limited by prohibitive on-device resource overhead, straggler-prone synchronous aggregation, and the assumption of a unified pretrained backbone across heterogeneous devices. In this paper, we propose EC-MobiLLM, a novel design for efficient collaborative LLM fine-tuning across heterogeneous mobile devices and heterogeneous backbones. EC-MobiLLM adopts a server-assisted side-tuning paradigm to minimize on-device overhead and pioneers non-blocking asynchronous collaboration to accelerate training. Moreover, EC-MobiLLM introduces adaptive feature alignment to support collaboration across heterogeneous backbones, aligning with practical needs in real-world deployments. Extensive experimental results demonstrate that EC-MobiLLM can maintain robust fine-tuning performance while achieving extremely low on-device memory, with at least 95.2\% reduction in computation overhead, 93.2\% reduction in communication costs and $5.1\times$ faster convergence compared to existing methods, validating its efficacy for practical LLM adaptation over heterogeneous mobile devices.

arXiv:2508.05941v2cs.RO

Latent Policy Barrier: Learning Robust Visuomotor Policies by Staying In-Distribution

Visuomotor policies trained via behavior cloning are vulnerable to covariate shift, where small deviations from expert trajectories can compound into failure. Common strategies to mitigate this issue involve expanding the training distribution through human-in-the-loop corrections or synthetic data augmentation. However, these approaches are often labor-intensive, rely on strong task assumptions, or compromise the quality of imitation. We introduce Latent Policy Barrier, a framework for robust visuomotor policy learning. Inspired by Control Barrier Functions, LPB treats the latent embeddings of expert demonstrations as an implicit barrier separating safe, in-distribution states from unsafe, out-of-distribution (OOD) ones. Our approach decouples the role of precise expert imitation and OOD recovery into two separate modules: a base diffusion policy solely on expert data, and a dynamics model trained on both expert and suboptimal policy rollout data. At inference time, the dynamics model predicts future latent states and optimizes them to stay within the expert distribution. Both simulated and real-world experiments show that LPB improves both policy robustness and data efficiency, enabling reliable manipulation from limited expert data and without additional human correction or annotation.

arXiv:2507.22017v5eess.IV

Cyst-X: A Multicenter MRI Benchmark for Pancreatic Cyst Risk Stratification and Cross-Institutional AI

Intraductal papillary mucinous neoplasms (IPMNs) are pancreatic cancer precursors, and guideline-based risk stratification both over-treats indolent lesions and misses progression. We present Cyst-X, a public benchmark of 1,461 T1-weighted (T1W) and T2-weighted (T2W) MRI volumes from 764 patients at seven centers, with expert pancreas masks, risk labels, splits and reference AI risk-prediction models. PanSegNet segmented the pancreas with Dice of 86.34% (T1W) and 89.26% (T2W) on held-out centers. For the primary task, high-risk versus low-risk IPMN, a 3D DenseNet-121 achieved cross-validated AUCs of 74.59% and 73.45%; complementary broad binary and three-class tasks reached 78.67--81.28% and 73.47--75.76%. Because patient images often cannot be pooled across institutions, we also simulated federated learning across the seven centers. On T2W MRI, FedProx was similar to centralized training in center-averaged external AUC (71.00% versus 71.18%) but not in pooled AUC (62.35% versus 77.49%). Radiologist comparisons showed cohort-dependent sensitivity--specificity tradeoffs. Cyst-X supports reproducible method development and identifies within-IPMN risk discrimination and cross-center score comparability as open challenges.

arXiv:2507.21740v2cs.NE

Knowledge-Guided Memetic Algorithm for Capacitated Arc Routing Problems with Time-Dependent Service Costs

The capacitated arc routing problem with time-dependent service costs (CARPTDSC) is a challenging combinatorial optimization problem that arises from real-world applications. CARPTDSC has two main challenges in reducing computational time. First, it is an NP-hard problem. Second, the time-dependent service costs of tasks require frequent evaluations during the search process, significantly increasing the computational burden. These challenges make it difficult for existing algorithms to perform efficient searches, often resulting in limited efficiency. To address these issues, this paper proposes a knowledge-guided memetic algorithm with golden section search and negatively correlated search (KGMA-GN), where two knowledge-guided strategies are introduced to improve efficiency. First, a knowledge-guided initialization strategy (KGIS) is proposed to generate high-quality initial solutions to speed up convergence. Second, a knowledge-guided small-step-size local search strategy (KGSLSS) is proposed to filter out invalid moves, thereby reducing unnecessary evaluations. Experimental results on five benchmark test sets, including both small- and larger-scale instances, demonstrate that KGMA-GN achieves higher search efficiency than the state-of-the-art methods. Moreover, the ablation study further confirms that the knowledge-guided local search operators in KGSLSS can significantly reduce runtime compared to traditional operators, especially for the knowledge-guided swap operator, which achieves more than a tenfold improvement in speed.

arXiv:2507.18229v3econ.GN

From Individual Learning to Market Equilibrium: Correcting Structural and Parametric Biases in RL Simulations of Economic Models

Applying Reinforcement Learning (RL) to economic modeling reveals a fundamental tension: standard RL agents do not behave as competitive price-takers. While macroeconomic models typically assume agents take aggregate conditions as given, a naive single-agent RL algorithm learns to internalize its impact on the environment, effectively exerting market power. We illustrate this divergence in a search-and-matching model with concave production, showing that a standard RL agent converges to a monopsonistic allocation rather than the competitive equilibrium. Furthermore, we show that RL's approach to time discounting distorts standard intertemporal economic trade-offs. To resolve these issues, we propose a Calibrated Mean-Field Reinforcement Learning framework that restricts the agent to optimize against an exogenous aggregate state to restore price-taking behavior, and recalibrates the RL reward structure to reflect true economic opportunity costs. By iteratively updating the aggregate state, our algorithm converges to a stationary fixed point that recovers the competitive equilibrium. This approach provides a method for embedding learning algorithms into dynamic general equilibrium models.

arXiv:2507.16271v3cs.CL

Beyond Isolated Dots: Benchmarking Structured Table Construction as Deep Knowledge Extraction

Large language models (LLMs) are increasingly expected to extract and synthesize structured information from complex real-world documents, yet they typically produce unstructured, hard-to-verify paragraph-style outputs. We introduce AOE (Arranged and Organized Extraction), a bilingual benchmark that evaluates LLMs' ability to distill long-form, multi-document inputs (64k to 1.8M tokens) into structured tables. AOE comprises 12 tasks across four domains (academic, financial, legal, and medical), each grounded in genuine analytical needs, with schema guidance provided to isolate extraction quality from formatting variance. We evaluate 12 frontier LLMs (both open-weight and proprietary) under four generation settings of increasing autonomy: direct full-document input, RAG-based retrieval, offline document-browsing agent, and online web-search agent. Even the best-performing model achieves only 42.7% Cell F1, and retrieval-augmented approaches often degrade rather than improve performance, revealing that structured extraction from long contexts remains a fundamental challenge - an "Area of Effect" (AOE) that impacts all tested systems.

arXiv:2507.10938v2cs.CV

Graph-Aggregated Prototype Learning for Remote Sensing Semantic Change Detection

Semantic change detection (SCD) extends the binary change detection task to provide not only the change locations but also the detailed "from-to" categories in multi-temporal remote sensing data. Such detailed semantic insights into changes offer considerable advantages for a wide array of applications. However, since SCD involves the simultaneous optimization of multiple tasks, the model is prone to negative transfer due to task-specific learning difficulties and conflicting gradient flows. To address this issue, we propose Graph-Aggregated Prototype Learning for Remote Sensing Semantic Change Detection(GAPL-SCD). In this framework, a multi-task joint optimization method is designed to optimize the primary task of semantic segmentation and change detection, along with the auxiliary task of graph aggregation prototype learning. Adaptive weight allocation and gradient rotation methods are used to alleviate the conflict between training tasks and improve multi-task learning capabilities. Specifically, the graph aggregation prototype learning module constructs an interaction graph using high-level features. Prototypes serve as class proxies, enabling category-level domain alignment across time points and reducing interference from irrelevant changes. Additionally, the proposed self-query multi-level feature interaction and bi-temporal feature fusion modules further enhance multi-scale feature representation, improving performance in complex scenes. Experimental results on the SECOND and Landsat-SCD datasets demonstrate that our method achieves state-of-the-art performance, with significant improvements in accuracy and robustness for SCD task.

arXiv:2507.01216v3cs.LG

Eco-MobiLLM: Efficient LLM Fine-Tuning on the Mobile Device via Server Assisted Side-Tuning

There is a huge gap between numerous intriguing applications fostered by on-device large language model (LLM) fine-tuning (FT) from fresh mobile data and the limited resources of a mobile device. While existing server-assisted methods (e.g., split learning or side-tuning) may enable LLM FT on the local mobile device, they suffer from heavy communication burdens of activation transmissions. In addition, their multiple on-device forward passes over the same samples across training epochs lead to substantial iterative computation on the mobile device. To address those issues, we develop Eco-MobiLLM, an efficient LLM FT method which can be deployed on the mobile device via server-assisted side-tuning. To reduce communication cost, Eco-MobiLLM develops an activation shortcut that transmits only the tokens involved in the loss calculation instead of full activation matrices to guide the side network tuning. Besides, to improve computing efficiency and further accelerate FT convergence, Eco-MobiLLM introduces a novel two-stage training paradigm, which allows the server to reuse historical activations and saves the mobile device from repeatedly computing forward passes for the recurring data samples. Extensive experimental results demonstrate Eco-MobiLLM's superiority.

arXiv:2506.22276v2cs.AI

Artificial Intelligent Disobedience: Rethinking the Agency of Our Artificial Teammates

Artificial intelligence has made remarkable strides in recent years, achieving superhuman performance across a wide range of tasks. Yet despite these advances, most cooperative AI systems remain rigidly obedient, designed to follow human instructions without question and conform to user expectations, even when doing so may be counterproductive or unsafe. This paper argues for expanding the agency of AI teammates to include \textit{intelligent disobedience}, empowering them to make meaningful and autonomous contributions within human-AI teams. It introduces a scale of AI agency levels and uses representative examples to highlight the importance and growing necessity of treating AI autonomy as an independent research focus in cooperative settings. The paper then explores how intelligent disobedience manifests across different autonomy levels and concludes by proposing initial boundaries and considerations for studying disobedience as a core capability of artificial agents.