PyTorchKR Papers

최신 논문

총 523,801편
태그·기간 등 상세 필터

카테고리 탐색

arXiv:2611.04027v1cs.RO

A Practical Causal Framework for Autonomous Mobile Robots: From Simulation to Real-World Human Environments

Causal inference has emerged as a powerful paradigm enabling robots to make safe, efficient decisions by reasoning about the ``why'' behind events. However, its real-world application is hindered by the difficulty of collecting causal data, especially in human-populated environments where training takes time, while users expect immediate operation. To this end, we tackle the challenge of transferring simulation-learnt causal relations to the real world with minimal effort. While retaining the invariant high-level causal structure, we empirically demonstrate the system adapts efficiently to real-world situations by updating only the causal model parameters with a limited set of new observations. Our results show that the parameter update enables rapid sim-to-real transfer. The framework runs efficiently onboard of a mobile robot and significantly outperforms a non-causal baselines navigating successfully and safely in complex human-shared scenarios.

arXiv:2611.04025v1eess.IV

DUNEDepth: Bronchoscopic Depth and Pose Estimation via Hybrid Adaptation of a 3D-Vision Foundation Model

State-of-the-art bronchoscopy guidance systems and computer-aided diagnosis methods have emerged as important tools for early lung cancer detection. Effective use of these tools depends on being able to accurately link a patient's bronchoscopic airway exam video to their pre-operative 3D chest CT scan. Unfortunately, the derivation of such links, which include bronchoscope depth and camera motion, from the exam video remains an open problem. Building on recently proposed 3D-vision foundation models, we propose DUNEDepth, a hybrid architecture that combines a shared DUNE ViT encoder, a twin MASt3R cross-attention decoder structure, and a prediction head that includes elements of the MASt3R and Reloc3r models. To fine-tune the architecture to our application, we utilize an integrated hybrid synthetic-supervised/self-supervised training strategy that draws on both bronchoscopic video and CT-derived virtual bronchoscopic views. Experiments drawing on bronchoscopic video sequences demonstrate that DUNEDepth more accurately derives depth information and camera motion trajectories than previously proposed self-supervised methods.

arXiv:2611.04012v1cs.CV

SurgEye: Instruction Tuning-Free Perception-Reasoning Framework for Surgical Scene Understanding

Surgical scene understanding is pivotal for computer-assisted intervention, yet demands compositional reasoning across instruments, anatomy, and workflow. Existing surgical vision-language models rely on instruction tuning with large-scale question-answer pairs, entailing laborious data curation and costly retraining. To bridge this gap, we propose SurgEye, a perception-reasoning framework that grounds reasoning in a vision foundation model without instruction tuning. Specifically, we devise Routed Multi-Task Perception (RMP) that routes the vision foundation model through task-specific pathways learned from public perception labels, producing spatial and temporal evidence. We further introduce Agentic Compositional Reasoning (ACR) that employs a frozen large language model to coordinate RMP specialists and temporal memory as callable tools, composing evidence across tasks and time into grounded answers. Moreover, we construct SurgEye-Bench spanning eight surgical procedures, pairing each question with perception labels for verifiable evaluation. Extensive experiments demonstrate that SurgEye outperforms state-of-the-arts in surgical reasoning. The source code will be available.

arXiv:2611.04010v1cs.LG

Decision Transformer with Task-Aware Representations for Offline Meta-Reinforcement Learning

To enable effective generalization in offline meta-RL, it is essential to accurately infer the underlying task through latent task representations, which provide a compact and transferable summary of task-specific characteristics for guiding policy generation. In this paper, we propose Contrastive Task-Aware Decision Transformer (CoTA-DT), a DT-based offline meta-RL framework that learns task representations by jointly integrating four complementary components: (i) contrastive learning from offline trajectories, (ii) alignment with natural-language task descriptions, (iii) one-step supervision via a task-conditioned world model, and (iv) long-term supervision via a task-conditioned Q function. The learned task representations are used as conditioning signals for a DT-based policy, enabling task-aware meta-policy learning. To further support generalization in both zero-shot and few-shot settings, we design a language-grounded, self-adaptive strategy that enables effective and reliable adaptation to unseen tasks. Notably, in CoTA-DT, the pretrained LLM serves as a shared backbone for language encoding, world modeling, Q-function estimation, and policy learning. This unified design facilitates parameter sharing, improving both efficiency and generalization. Extensive experiments across diverse offline meta-RL benchmarks demonstrate that CoTA-DT consistently outperforms prior methods, particularly in challenging zero-shot generalization scenarios.

arXiv:2611.04007v1cs.LG

Physics-Informed Identification of Tensegrity Prestress across Taut-Slack Transitions

Prestress governs the stability, stiffness, and dynamic behavior of tensegrity structures, but it is not directly observable from sparse structural measurements, and it may decrease during service life. Under strong excitation, the cables may slacken and the dynamics become nonsmooth. Existing identification methods are limited to the taut regime, and physics-informed networks for tensegrities solve only the forward problem of predicting the response from known member parameters. This study formulates the inverse problem. Neural networks are trained on the equation-of-motion residual in constrained Cartesian coordinates, where the mass matrix is constant and the force densities enter linearly. The force densities, damping, and axial stiffness are trainable parameters, and the tension-only cable law is kept. A Fisher-information analysis shows, before inversion, which member parameters a record can determine. The method is tested on simulated planar tensegrity frames with up to five modules and a three-dimensional prism, and on measured records of a biomimetic leg and a full-scale tensegrity simplex. On the one-module frame, the identifiable force densities are recovered within 1.5% from noisy accelerations at half the free coordinates, on records with and without taut-slack transitions. Records with transitions determine members that taut records leave undetermined. On the five-module frame, a 16-cable two-span bridge, the median force-density error rises moderately, to 4.2%. On measured records, an identified spring stiffness is within 0.1% of its independently measured value on average, and the identified simplex prestress at 13 levels agrees with its cable force transducers to 2.9% root-mean-square.

arXiv:2611.04006v1cs.AI

Controlling Action-Head Co-Adaptation to VLM Representation Anisotropy

Modern learning systems increasingly combine pretrained models with downstream heads. Despite the growing importance of such modular designs, how the downstream model co-adapts to the geometry of pretrained representations remains relatively underexplored. We study this problem through vision-language-action models (VLAs), where a pretrained vision-language model (VLM) conditions an action model for control. We find that the action model progressively concentrates its sensitivity on high-variance dimensions of the VLM outputs. Through controlled experiments across different backbone-head models, we further show that counteracting this sensitivity concentration can be beneficial under representation--task mismatch settings. Motivated by these observations, we introduce \emph{Adaptive Interface Equalizer} (AdaIE), a parameter-free mechanism that reweights pretrained representations according to the evolving downstream sensitivity during training. AdaIE improves VLA success rates on simulation benchmarks, with gains of up to $+5.28$ percentage points (pp) on RoboCasa-GR1 and $+1.68$ pp on RoboTwin 2.0, and achieves an average gain of $+19$ pp across five real-world manipulation tasks. It further generalizes to diffusion policies and world models, improving downstream performance in both settings. Our results highlight interface co-adaptation as an important optimization problem when coupling pretrained representations with downstream models.

arXiv:2611.04005v1cs.IR

Characterizing Adversarial GEO via Implicit Brand Injection

Generative search engines increasingly shape how users discover products, services, and commercial entities by generating answers from retrieved web content. This creates an opportunity to influence which brands appear in these answers by modifying the content available to the retrieval system. In this work, we study an adversarial generative engine optimization (GEO) setting in which a predefined brand is added to an article through a single contextually integrated mention. The challenge is to make the mention fit the surrounding article without making the content overtly promotional, while keeping it relevant enough to be retrieved and to influence the generated answer. To study this, we introduce Implicit Brand Injection (IBI), a multi-agent framework that identifies insertion locations, rewrites the selected sentence, verifies the edit, and selects the final candidate. To evaluate IBI at scale, we construct IBIBench, which contains 1,000 web articles and 60 synthetic brands across 12 diverse domains, with both endemic and non-endemic brand placements. We compare IBI with Direct Prompting and Best-of-N with ranking under an equivalent budget, evaluating insertion quality, target-chunk retrieval, and downstream brand propagation. Compared with the baselines, IBI produces insertions that better preserve semantics and disrupt perplexity less, while improving Hit@k across all evaluated cutoffs, query regimes, and placement settings. IBI achieves up to 82.2% Hit@10 and 0.74 MRR@10, while Brand Mention Rate (BMR) and Conditional Propagation (CP) reach up to 62.7% and 75.9%, respectively. These results show that a single brand mention can increase both retrieval exposure and brand visibility in generated responses, highlighting the need for defenses against implicit content manipulation in generative search.

arXiv:2611.04003v1cs.AI

WASP-Mem: Let Agent Organize Procedural Memory. Human Spend Effort on Diagnosis

LLM agents struggle to automate complex enterprise procedures because static, pre-built harness cannot capture proprietary policies, undocumented behaviors, or state-dependent recovery paths that surface only during execution. Procedural memory accrued from agent experience offers a training-free path to closing this gap -- yet unlike episodic records or semantic facts, it must consolidate repeated successes and diagnosed failures into reusable actions and recovery strategies. We introduce WASP-Mem (Wiki-based Agent-Synthesized Procedural Memory), a nimble but effective method that distills execution traces and feedback into persistent, interlinked wiki pages, enabling continual agent learning. On TheAgentCompany benchmark, WASP-Mem raises the checkpoint pass rate by 4.6-6.1% over the strongest baseline we compared across three models while requiring 57-61% less storage. We further present a systematic controlled study spanning consolidation method, feedback type, trace ingestion scope, and refinement rounds. Our results reveal that prescriptive knowledge organization such as human defined schemas and explicit deduplication rules hurts performance rather than closing the procedural-knowledge gap. In contrast, human diagnosis converts failed traces into corrective knowledge and drives sustained improvement: a single round of human diagnosis outperforms four rounds of self-judgment by 7.0-10.9%. These findings point to a simple division of labor: let agents organize procedural memory, and invest human effort where it matters most -- diagnosis.

arXiv:2611.04002v1eess.IV

Evaluating the Impact of Nonlinear Registration on Longitudinal PET Analysis in a Rodent Stroke Model

Longitudinal rodent positron emission tomography (PET) studies of stroke and neuromodulation rely on accurate registration to a common atlas space to detect group-level effects. Prior work applying ultra-high frequency low-intensity magnetic stimulation (UHF-LiMS) to a rat model of chronic capsule infarct used affine normalization to a Sprague Dawley MRI template without evaluating that choice. We compare a two-stage (affine and nonlinear) pipeline against an affine- only one for rat PET-to-atlas alignment on the same dataset. Across 32 longitudinal [18F]-FDG PET scans (8 rats, 4 timepoints), the nonlinear stage gave a higher Dice similarity coefficient on every scan, and the gain was significant at the rat level. Jacobian determinants showed no folding on the sampling grid. Group-level linear mixed-effects maps of change from baseline (AFNI 3dLME), computed identically under both pipelines, differ in the size, number and location of their suprathreshold clusters. In this dataset, the registration pipeline is therefore a variable that can change the interpretation of downstream group-level analyses.

arXiv:2611.03998v1cs.CL

Dory1.0-4B: A Looped Normalized-GPT

We introduce Dory1.0-4B, a 4-billion-parameter looped (latent-recurrent) language model with elastic depth: a single checkpoint serves different inference compute budgets. Dory1.0-4B is built on a normalized Transformer (nGPT) backbone with interleaved sliding-window and full-attention layers, organized into an input block, a shared recurrent block, and an output block. At inference, the recurrent block can be applied a variable number of times. We pre-trained Dory1.0-4B on 6 trillion tokens. The first 4.5 trillion tokens were trained in NVFP4, and additional pre-training then extended the context length to 256K tokens. Next, we post-trained the model using Supervised Fine-Tuning (SFT) and Reinforcement Learning with Verifiable Rewards (RLVR), with a difficulty-aware loop curriculum that routes harder prompts to deeper loops. Increasing the loop count improves accuracy, with the largest gains on coding, agentic, and long-context benchmarks. With two or more loops, Dory1.0-4B outperforms larger models such as Qwen3.5-9B, Gemma 4 12B, and Gemma 4 26B-A4B on average across 12 benchmarks. Looping also makes Dory1.0-4B more token-efficient on agentic benchmarks: with four loops, it generates 1.3-2.5x fewer tokens than with one loop, as the extra computation is spent in latent space rather than on longer outputs. To reduce memory during looping, Dory1.0-4B caches keys and values only from the last loop, keeping its KV-cache footprint equal to that of a non-looped model at any loop count while retaining most of the accuracy gains. We open-source the BF16, MXFP8, NVFP4, and GGUF quantized checkpoints.

arXiv:2611.03996v1eess.IV

Measure Before You Repair: An Explainable Agent for Blind Restoration of Multi-Artifact 3-D MRI

A brain MRI scan can carry an unknown mix of artifacts, so restoration has to be blind, yet most deep-learning methods are built to correct one known artifact. We present Dx-Rx, an agent in which a multimodal large language model (MLLM) measures suspected artifacts before it repairs them. To our knowledge, it is the first agent to restore whole 3-D MRI volumes, whereas earlier MRI agents restore 2-D slices. In a loop, it looks at slices of the 3-D volume, calls five diagnostic tools (noise, spike, ghosting, bias field, motion) and four restoration tools in any order it chooses, checks each repair, can return to any earlier version to change the order, and ends with a radiology-style report of its diagnostic reasons, the repairs it applied and their results. We simulated up to five artifacts on 148 whole 3-D T1-weighted volumes, at levels calibrated to radiologists' grades of real scans. An MLLM that only looked at the image, as earlier MRI agents do, removed at most 19.7 % of the artifact error. With the best of five MLLMs, Dx-Rx removed a mean 64.5 %, nearly matching the 64.6 % of the same tools given the ground-truth artifacts, which are never known for a clinical scan. With all five models, it made no volume worse and was on par with or better than an expert rule-based pipeline built from the same tools (62.3 %). Code and demo are public at https://itu-biai.github.io/measure-before-you-repair.

arXiv:2611.03995v1eess.SY

When Music Needs Meaning: Semantic Communication as a Transformative Enabler for the Musical Metaverse

The Musical Metaverse (MM) enables geographically distributed musicians, audiences, sound engineers, and AI agents to interact through synchronized audio, visual, kinematic, and haptic streams, imposing latency, synchronization, bandwidth, and scalability requirements that are difficult to satisfy with conventional bit-level communication. This paper argues that semantic communication, understood here as the transmission of musically meaningful information rather than raw audio-visual-haptic samples, provides a principled communication paradigm for the MM. We define the MM from a communication perspective and derive modality-specific require ments; propose a three-pillar framework for musical semantic communication comprising semantics extraction, semantic-aware orchestration, and cross-modal semantic synchronization; survey representative work across semantic communication, music AI and edge orchestration, identifying the gaps that prevent their integration; and propose a semantic-first reference architecture based on device-side semantic encoders, edge semantic routers, and cross-modal semantic synchronization. The resulting framework positions the MM as a concrete and demanding use case for 6G semantic communication.

arXiv:2611.03993v1cs.CL

Whose startup does the AI back? An algorithm audit of reputation and demographic signals in LLM-assisted crowdfunding

Large language model (LLM) assistants increasingly advise backers on which crowdfunding campaigns to support. LLMs have known race and gender problems, and it is unsettled whether they reproduce, erase, or reverse the discrimination documented in entrepreneurial finance. We ask when, and through which decision process, LLM assistants use founders' gender and ethnicity. In a pre-specified choice-based conjoint experiment, seven LLMs chose among five synthetic campaigns with independently randomized quality signals and founder names, under three backer personas and a persona-free control (40,248 responses). By default, founder names barely matter: under budget-conscious and delivery-focused personas, gender and Black-name effects are statistically equivalent to zero. When the backer wants to support underrepresented creators, the models favor Black-signaled (+8.1 percentage points; positive in all seven models) and female-signaled (+5.1) founders. They do so most strongly among near-tied campaigns and for the founder who alone carries the signal, and they almost never name gender or ethnicity in their reasons. The results fit an agency account of algorithmic discrimination: an aligned assistant takes its demographic preferences from its user rather than from its training data.

arXiv:2611.03992v1cs.CL

FlagRot: Selective Rotation for Low-Bit KV Caches

Long-context inference is bound by the KV cache, which 2-bit codes could shrink about sixfold if their error stays in check. The usual remedy rotates keys before quantizing, assuming rotation always helps. However, we observe that it does not: rotation flattens a spiky piece of a key but can give a flat one a new peak, and both coexist within one key. We therefore propose FlagRot, which stores each 32-channel piece rotated or as is, whichever has the smaller calibrated attention error, and rotates the query instead when reading. We further show that how often to decide follows the quantization axis: per-channel caches such as KIVI need one decision per head, at no write cost, while per-token caches decide per key, with the choice stored in existing sign bits. At 2.5 bits per value, FlagRot lowers the twelve-model KL divergence of KIVI from 18.7 to 12.4 x 10^-3, below RotateKV on 9 of 12 models. Served as a vLLM plugin, it raises RULER at 128k tokens by +2.9 points and the LongBench v1 mean by +0.4 to +1.9 points on the three of four models where 2-bit KIVI loses accuracy. It also lowers the error of 4-bit integer and FP4 caches.

arXiv:2611.03985v1cs.AI

Thinking Fast, Trading Late: System One Decision Models vs the Market

Large language models (LLMs) read company news at scale, and backtests suggest that their readings predict stock returns. We ask how much of that predictability survives a test that rules out look-ahead, memorization and selection. We propose OLIVIA, a protocol for such tests: point-in-time inputs with company identities masked, selection on validation data only, an analysis plan frozen by cryptographic hash, and a test period unblinded once through a logged lock. We apply it to AI decision models, non-generative language models that return probabilities for typed questions, reading news about S&P 500 stocks: an open-weights model we fine-tuned on 2024-2025 data, three commercial models and standard baselines. The fine-tuned model's signal replicated on two validation quarters (daily information coefficients 0.055 and 0.065) but fell to 0.020 (t = 1.7) on the blinded January-July 2026 test. The commercial models' readings line up with the overnight move to the open (information coefficient 0.074 to 0.077) but not with the move after it, so the news is priced by the time a trader can act on it at the open. No model can tell over- from under-reaction, and no pre-registered strategy survives costs and a deflated Sharpe ratio. Commercial models identify masked companies in 62% to 74% of cases against a 28% baseline: masking hides names, not companies. The market reads the news first.

arXiv:2611.03983v1cs.AI

A Guideline-Grounded Multi-Agent Framework for Personalized Exercise Prescription from Clinical and Wearable Data

Translating population-level exercise guidance into individualized prescriptions requires integrating clinical and wearable data, including diagnoses, medications, functional limitations, and baseline activity, with condition-specific recommendations, a process difficult to scale beyond specialist rehabilitation settings. We developed FITTER, a multi-agent framework that integrates these data with condition-specific guidance to generate personalized FITT-VP prescriptions and applies deterministic safety audit and repair to correct rule-checkable violations. We evaluated FITTER using data from 400 All of Us participants (300 cancer, 100 stroke) against five LLM comparators through computational evaluation, ablation, and structured clinician review. FITTER achieved 98.5% safety-constraint compliance versus 75.8-85.8% for comparators and showed greater quantitative dosing precision and guideline alignment. Cancer-cohort ablation identified safety audit and repair as the largest contributor to safety-constraint compliance. Clinicians reviewed 80 prescriptions across safety, personalization, instruction concreteness, anticipated effectiveness, and clinical acceptability. FITTER provides a clinician-facing approach integrating clinical and wearable data, guidance, and explicit safety constraints.

arXiv:2611.03982v1cs.CL

SLATE: A Large-Scale Benchmark for Long-Horizon API Selection in E-commerce Tool Use

Enterprises increasingly want LLM agents to automate standard operating procedures over their software stacks -- mapping each step of a procedure to the right call among thousands of internal APIs, without hand-authored step-to-API mappings. The core difficulty is API selection at scale: choosing and composing the correct tools over a long horizon, judged only by whether the overall task succeeds. Current tool-use benchmarks do not test this regime -- they expose only a handful of tools, evaluate single-step invocation, or grade multi-step behavior with subjective LLM-as-a-judge scores that penalize valid alternative trajectories. We introduce SLATE (Synthetic Large-scale API Toolkit for E-commerce), a large-scale benchmark that tests large-toolset, long-horizon API selection in a realistic enterprise domain. SLATE couples (i) queries decomposed into hierarchical, tool-grounded procedures over a ~1,000-tool e-commerce library, (ii) a deterministic, context-aware simulator that returns recorded outputs for matching calls and default responses otherwise, and (iii) an end-to-end task-success metric -- a task succeeds when it reaches every reference outcome -- that credits diverse-but-valid trajectories while remaining fully automatic and reproducible. Experiments with six open decision models and the language models they are built on, from 3B to 12B parameters, show that even the best model completes only 5.0% of the tasks, establishing SLATE as a challenging testbed for reliable enterprise tool use. We release SLATE, its simulator, and the evaluation harness at https://geshijoker.github.io/slate/

arXiv:2611.03981v1cs.LG

Rapid Mean-Field Formation Transport with a Shared-Representation Neural Fokker-Planck Controller

This paper develops a mean-field formation transport method with low online computational cost for large-scale multi-agent systems. Classical mean-field planners often require repeated solutions of coupled forward and backward partial differential equations, making online replanning computationally demanding. Density and control are therefore approximated by a common parameterized nonlinear map with separate scalar density and vector control outputs. Parameters are optimized using Fokker--Planck and initial-condition residuals, formation-matching costs on numerically simulated closed-loop particle trajectories, and control effort. Numerical computations are restricted to a bounded region of the state space, with an additional penalty on probability flux across its boundary. For the periodic-domain model, we establish well-posedness of the induced density evolution and bound the density-approximation error in terms of the dynamics and initial-condition residuals. Parameter optimization precedes online operation, during which controller coefficients remain fixed. Experiments quantify the accuracy--runtime trade-off relative to a mean-field planner, show lower transport error than classical feedback baselines, and demonstrate the benefits of the common parameterization.

arXiv:2611.03980v1cs.AI

Measuring Model-Specific Alignment Drift in LLM-Based Opinion Dynamics

Large language model (LLM) agents are increasingly used as synthetic participants in opinion-dynamics experiments, often assuming that an agent's behaviour is fully determined by its prompt. We test this assumption with a framework in which LLM agents discuss a contested policy issue over repeated rounds while their stance is quantified by a fixed psychometric instrument independent of any dynamics model. Across six LLMs, committed personas are intrinsically stable: in each of the five models tested without peer input, every agent keeps its assigned side for 20 rounds. The observed loss of alignment under discussion is therefore interaction-induced, we call it alignment drift, and model-specific: gpt-4.1 keeps opinion clusters apart, claude-sonnet-5 holds them partially, and gpt-5.6-sol and claude-opus-5 abandon them. Under selective attention, one LLM maintains separated opinion clusters under stance-homophilous attention. Qualitative profiles persist at doubled population and horizon and on a second issue. Such heterogeneity makes the LLM choice a first-class validity variable for social simulation.

arXiv:2611.03976v1cs.HC

Must AI Spell It Out? Pragmatic Comprehension of Alerts in a Time-Critical Task

Assistive systems typically tell users everything, often spending time and attention on what a user already knows, leaving less for what happens next. People, on the other hand, leave out what they expect the listener can supply. Can people use their prior knowledge to understand an incomplete message when the speaker is a machine though? We studied this pragmatic comprehension in a monitoring task controlling users' prior knowledge and the available time to detect critical situations. In a preregistered study (N=114), we compared a pragmatic planner that reasons over a model of the listener's knowledge to decide what to say, when, and how explicitly, with a baseline specifying problems fully in onset order. Participants resolved underspecified alerts almost as reliably as fully specified ones and sooner, detected more problems, and rated the pragmatic assistant helpful more often, with no increase in reported mental demand. Our findings show that people can recover what the system left implicit and that pragmatic communication can support effective assistance under the constraints of a changing world.