NVIDIA、Nemotron Speechで臨床ASRを高速化

Operadic consistency: a label-free signal for compositional reasoning failures in LLMs

arXiv cs.CL (Computation and Language) ・ 2026-06-11

LLMの合成推論の失敗をラベル不要で検出する「オペラド整合性」を提案

Valid Inference with Synthetic Data via Task Exchangeability

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-11

合成データで妥当な推論を保証するtask exchangeabilityを提案

Beyond Uniform Tokens: Adaptive Compression for Time Series Language Models

arXiv cs.CL (Computation and Language) ・ 2026-06-11

時系列LLMの非対称トークン圧縮で効率化、予測から異常検知まで有効

Beyond the Commitment Boundary: Probing Epiphenomenal Chain-of-Thought in Large Reasoning Models

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-11

思考連鎖に『コミットメント境界』、以降の手順は無影響と判明

Simplex-Constrained Sparse Bagging: Transitioning from Uniform Priors to Sparse Posteriors in Ensemble Learning

arXiv cs.LG (Machine Learning) ・ 2026-06-11

単体制約スパースバギングSCSB、アンサンブルを最大96%圧縮し較正改善

Existence Precedes Value: Joint Modeling of Observational Existence and Evolving States in Time Series Forecasting

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-11

Timeflies、未来の観測有無と値を同時推定する予測枠組み

A2D2: Fine-Tuning Any-Length Discrete Diffusion for Adaptive Decoding

arXiv cs.LG (Machine Learning) ・ 2026-06-11

A2D2、可変長離散拡散モデルの報酬誘導ファインチューニングを統一

Is It You or Your Environment? A Bayesian Inference Framework for Genomically-Anchored Personalized Physiological Interpretation

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-11

ゲノムを事前分布に使い個別生理解釈の冷却開始を解決

NetCause: Counterfactual Learning for Root Cause Analysis in Large-Scale Networks

arXiv cs.LG (Machine Learning) ・ 2026-06-11

NetCause、反実仮想学習でクラウド網障害の根本原因を順位付け

Graphical Causal Reasoning for Root Cause Analysis in Cloud Networks

arXiv cs.LG (Machine Learning) ・ 2026-06-11

クラウド網障害の根本原因分析、因果グラフ探索で85.7%の再現率

Heterogeneous LiDAR Early Fusion and Learned Re-Ranking Strategy for Robust Long-Term Place Recognition in Unstructured Environments

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-11

異種LiDAR早期融合と再ランクで非構造環境の場所認識を頑健化

GF-DiT: Scheduling Parallelism for Diffusion Transformer Serving

arXiv cs.LG (Machine Learning) ・ 2026-06-11

GF-DiT、拡散Transformer配信のGPU並列度を動的スケジューリング

Optical Implementation of Equilibrium Propagation Using Spatial Photonic Ising Machines

arXiv cs.LG (Machine Learning) ・ 2026-06-11

空間フォトニックイジングマシンで平衡伝播を光学実装

Accelerating Speculative Diffusions via Block Verification

arXiv cs.LG (Machine Learning) ・ 2026-06-11

拡散モデルに投機的復号のブロック検証を導入、受理率を理論的に改善

PolyFlow: Safe and Efficient Polytope-Constrained Flow Matching with Constraint Embedding and Projection-free Update

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-11

PolyFlow、多面体制約を埋込む射影不要のflow matching

MiniMax Sparse Attention

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-11

MiniMax、超長文脈向けブロック疎注意MSAを発表

SmartFont: Dynamic Condition Allocation for Few-Shot Font Generation

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-11

SmartFont、大域と局所条件を多段配分する少数事例フォント生成

Hölder++: Improving the Quality-Coherence Trade-off in Multimodal VAEs

arXiv cs.LG (Machine Learning) ・ 2026-06-11

Hölder++、マルチモーダルVAEの品質と整合性の両立を改善

VideoMDM: Towards 3D Human Motion Generation From 2D Supervision

arXiv cs.LG (Machine Learning) ・ 2026-06-11

VideoMDM、3D正解データなしで動画の2D姿勢から3D動作生成を学習

SkillCAT: Contrastive Assessment and Topology-Aware Skill Self-Evolution for LLM Agents

arXiv cs.CL (Computation and Language) ・ 2026-06-11

SkillCAT、LLMエージェントのスキル自己進化を検証付き3段階に分離

SICI: A Semantic-Pragmatic Complexity Index Reveals Regime Shifts in LLM Stance Detection

arXiv cs.CL (Computation and Language) ・ 2026-06-11

意味・語用論的複雑性指標SICI、LLM立場検出のレジーム変化を解明

MiniPIC: Flexible Position-Independent Caching in <100LOC

arXiv cs.CL (Computation and Language) ・ 2026-06-11

MiniPIC、100行未満でvLLMに位置非依存KVキャッシュを実装

Reroute, Don't Remove: Recoverable Visual Token Routing for Vision-Language Models

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-10

Reroute、視覚トークンを除去でなく再ルーティングしVLM推論を効率化

Context-Driven Incremental Compression for Multi-Turn Dialogue Generation

arXiv cs.CL (Computation and Language) ・ 2026-06-10

C-DIC、多ターン対話の文脈を逐次圧縮し長対話を安定化

DIRECT: When and Where Should You Allocate Test-Time Compute in Embodied Planners?

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-10

DIRECT、身体エージェントのテスト時計算をプロンプト別に配分

Doc-to-Atom: Learning to Compile and Compose Memory Atoms

arXiv cs.CL (Computation and Language) ・ 2026-06-10

Doc2Atom、文書を知識アトムに分解し合成的パラメトリック記憶を構築

System Report for CCL25-Eval Task 5: New Dataset and LoRA-Fine-Tuned Qwen2.5

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-10

PoetryQwen、CCPoetry-49Kで古典中国詩の鑑賞を専門化

TAHOE: Text-to-SQL with Automated Hint Optimization from Experience

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-10

Tahoe、経験からヒントを学びText-to-SQLを本番最適化

ATLAS: Active Theory Learning for Automated Science

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-10

ATLAS、能動学習で解釈可能な行動モデルを自動的に発見

APPO: Agentic Procedural Policy Optimization

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-10

APPO、細粒度の決定点で分岐と信用割当を行うエージェントRL

Breaking Entropy Bounds: Accelerating RL Training via MTP with Rejection Sampling

arXiv cs.CL (Computation and Language) ・ 2026-06-10

Bebop、棄却サンプリングでMTP受理率を改善しRL学習を加速

Latent World Recovery for Multimodal Learning with Missing Modalities

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-10

LWR、欠損モダリティ下で潜在世界を復元しマルチモーダル学習

CHORUS: Decentralized Multi-Embodiment Collaboration with One VLA Policy

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-10

CHORUS、単一VLA方策で分散的な多体協調を実現

Claw-SWE-Bench: A Benchmark for Evaluating OpenClaw-style Agent Harnesses on Coding Tasks

arXiv cs.CL (Computation and Language) ・ 2026-06-10

Claw-SWE-Bench、OpenClaw型エージェントのコード能力を多言語評価

ALIGNBEAM : Inference-Time Alignment Transfer via Cross-Vocabulary Logit Mixing

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-10

ALIGNBEAM、語彙を越えて安全アンカーのlogitを移植

Fourier Features Let Agents Learn High Precision Policies with Imitation Learning

arXiv cs.LG (Machine Learning) ・ 2026-06-10

Fourier 特徴で点群方策が高精度ロボット操作を獲得

Measuring Semantic Progress in Multi-turn Dialogue via Information Gain

arXiv cs.CL (Computation and Language) ・ 2026-06-10

多ターン対話の意味的進展を情報利得で測る指標を提案

PROJECTMEM: A Local-First, Event-Sourced Memory and Judgment Layer for AI Coding Agents

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-10

projectmem、コーディングエージェントに局所優先の記憶層を追加

A Five-Plane Reference Architecture for Runtime Governance of Production AI Agents

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-10

本番AIエージェントの実行時統治に5プレーン参照アーキテクチャ

Harness In-Context Operator Learning with Chain of Operators

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-10

CHOP、演算子連鎖で凍結ICONを分布外タスクへ汎化

CCKS: Consensus-based Communication and Knowledge Sharing

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-10

CCKS、合意ベースの通信と知識共有で協調的MARLを改善

Holding the FP8 Quality Ceiling at 8-Bit Weights and Activations: INT8 and GGUF Post-Training Quantization of Ideogram 4.0 for Consumer GPUs

arXiv cs.LG (Machine Learning) ・ 2026-06-10

Ideogram 4.0 を INT8 量子化、民生 GPU で FP8 品質を維持

Mathematical perspective on genetic algorithms with optimization guided operators

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-10

最適化誘導演算子を持つ遺伝的アルゴリズムの数理的視座

The Impossibility of Eliciting Latent Knowledge

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-10

潜在知識の引き出し(ELK)は不可能と因果影響図で形式化

VIA-SD: Verification via Intra-Model Routing for Speculative Decoding

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-10

VIA-SD、モデル内ルーティングで投機的デコードを高速化

Re-evaluating Confidence Remasking in Masked Diffusion Language Models

arXiv cs.LG (Machine Learning) ・ 2026-06-10

拡散言語モデルの remasking 手法 WINO、再評価で効果は限定的

Can News Predict the Market? Limits of Zero-Shot Financial NLP and the Role of Explainable AI

arXiv cs.CL (Computation and Language) ・ 2026-06-10

ゼロショット金融NLPの限界、ニュースで株価予測は困難と示す

Intelligent Automation for Embodied Benchmark Construction: Pipelines, Embodiments, Simulators, and Trends

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-10

身体ベンチ構築の自動化を5段階パイプラインで調査

Adaptive Multi-Resolution Procedural Knowledge Compression for Large Language Models

arXiv cs.CL (Computation and Language) ・ 2026-06-10

SKIM、手続き的スキルを適応的多解像度で圧縮しコスト削減

Implicit Neural Representations of Individual Behavior

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-10

Behavioral INR、無ラベルの多方策行動から方策表現を学習

Which Speech Representation Better Matches Text-Native Reasoning? A Study of Speech-Text Alignment on Frame Rate and Representation

arXiv cs.CL (Computation and Language) ・ 2026-06-10

音声表現とテキスト的推論の整合をフレームレート観点で研究

Agentic Environment Engineering for Large Language Models: A Survey of Environment Modeling, Synthesis, Evaluation, and Application

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-10

エージェント環境工学を環境のライフサイクル観点で調査

A Resource for Enthymeme Detection in Controversial Political Discourse

arXiv cs.CL (Computation and Language) ・ 2026-06-10

論争的政治言説のエンチメーム検出データセットを公開

Towards Responsibly Non-Compliant Machines

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-10

責任ある「不服従」が可能なAIエージェントの工学を提起

FORT-Searcher: Synthesizing Shortcut-Resistant Search Tasks for Training Deep Search Agents

arXiv cs.CL (Computation and Language) ・ 2026-06-10

FORT、近道耐性のある探索課題を合成し深層探索エージェントを訓練

"That's AI Slop, You Bot!" Studying Accusations, Evidence, and Credibility in Online Discourse Towards LLM-Generated Comments

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-10

「AIスロップ」非難の急増を2500万コメントで分析

Toward Generalist Autonomous Research via Hypothesis-Tree Refinement

arXiv cs.CL (Computation and Language) ・ 2026-06-10

Arbor、仮説ツリー改良で長期間の自律研究ループを運用

When Does Language Matter? Multilingual Instructions Reveal Step-wise Language Sensitivity in Vision-Language-Action Models

arXiv cs.CL (Computation and Language) ・ 2026-06-10

VLAモデルの言語頑健性は段階ごとの制御問題と判明

Notes2Skills: From Lab Notebooks to Certainty-Aware Scientific Agent Skills

arXiv cs.CL (Computation and Language) ・ 2026-06-10

Notes2Skills、実験ノートを確信度付きの科学エージェントスキルに変換

Beyond representational alignment with brain-guided language models for robust reasoning

arXiv cs.CL (Computation and Language) ・ 2026-06-10

脳誘導の言語モデルで頑健な演繹推論を強化

Fine-tuning Multi-modal LLMs with ART: Art-based Reinforcement Training

arXiv cs.CL (Computation and Language) ・ 2026-06-10

ART、視覚入力のみ最適化で凍結MLLMをソフトトークン微調整

WorldReasoner: Evaluating Whether Language Model Agents Forecast Events with Valid Reasoning

arXiv cs.CL (Computation and Language) ・ 2026-06-10

WorldReasoner、エージェントの事象予測の推論妥当性を評価

MultiToP: Learning to Patch Visual Tokens to Mitigate Hallucinations in Video Large Multimodal Models

arXiv cs.CL (Computation and Language) ・ 2026-06-10

MultiToP、視覚トークンを修復し動画LMMの幻覚を緩和

Fast Speech Foundation Model Distillation Using Interleaved Stacking

arXiv cs.CL (Computation and Language) ・ 2026-06-10

交互スタッキングで音声基盤モデルの蒸留学習を高速化

EEVEE: Towards Test-time Prompt Learning in the Real World for Self-Improving Agents

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-09

EEVEE、複数データセット対応のテスト時プロンプト学習を実現

Data Journalist Agent: Transforming Data into Verifiable Multimodal Stories

arXiv cs.CL (Computation and Language) ・ 2026-06-09

Data2Story、データから検証可能なマルチモーダル記事を自動生成

Multi-Faceted Interactivity Alignment in Full-Duplex Speech Models

arXiv cs.CL (Computation and Language) ・ 2026-06-09

全二重音声対話モデルの対話性をRLで多面的に改善

ReasonAlloc: Hierarchical Decoding-Time KV Cache Budget Allocation for Reasoning Models

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-09

ReasonAlloc、推論モデルのKVキャッシュ予算を階層的に配分

Itô maps for any-step SDEs

arXiv cs.LG (Machine Learning) ・ 2026-06-09

Itoマップ、確率動力学の任意ステップ流れマップを単一パスで予測

ABC-Bench: An Agentic Bio-Capabilities Benchmark for Biosecurity

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-09

ABC-Bench、LLMエージェントのバイオセキュリティ関連能力を評価

FADA: Accessible fetal ultrasound interpretation and annotation with a selectively distilled unified vision-language model

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-09

FADA、単一の視覚言語モデルで胎児超音波を統合解釈

RoboNaldo: Accurate, Stable and Powerful Humanoid Soccer Shooting via Motion-Guided Curriculum Reinforcement Learning

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-09

RoboNaldo、動作誘導カリキュラムRLでヒューマノイドの強シュート

A History-Aware Visually Grounded Critic for Computer Use Agents

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-09

HiViG、計算機操作エージェント向けの履歴認識・視覚接地批評

T1-Bench: Benchmarking Multi-Scenario Agents in Real-World Domains

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-09

T1-Bench、25ドメインの多シナリオエージェントを高忠実評価

What Fits (Into Few Tokens) Doesn't Overfit: Compression and Generalization in ML Research Agents

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-09

ML研究エージェントの圧縮性が過学習の少なさを説明

Workflow-GYM: Towards Long-Horizon Evaluation of Computer-use Agentic tasks in Real-World Professional Fields

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-09

Workflow-GYM、専門ソフトの長期GUIタスクを評価

AuRA: Internalizing Audio Understanding into LLMs as LoRA

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-09

AuRA、音声理解をLoRAとしてLLMに内在化

Diffusion Forcing Planner: History-Annealed Planning with Time-Dependent Guidance for Autonomous Driving

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-09

DFP、履歴誘導の拡散計画で自動運転の軌跡を安定化

Understanding and mitigating the risks of OpenClaw for non-technical users: A practical guide with Skill

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-09

OpenClawの危険を非専門ユーザ向けに7分類し平易に解説

Mind the Gap: Can Frontier LLMs Pass a Standardized Office Proficiency Exam?

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-09

最先端LLM、Office操作試験で最高36.6%にとどまる

CLP: Collocation-Length Prediction for Zero-Loss Adaptive Multi-Token Inference

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-09

CLP、バックボーンを設計者とし零損失の適応的多トークン推論

Frontier Coding Agents Use Metaprogramming to Adapt to Unfamiliar Programming Languages

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-09

最先端コーディングエージェント、難解言語へメタプログラミングで適応

Role-Agent: Bootstrapping LLM Agents via Dual-Role Evolution

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-09

Role-Agent、単一LLMがエージェントと環境を兼ね共進化

Range Penalization: Theoretical Insights with Applications in Federated Learning

arXiv cs.LG (Machine Learning) ・ 2026-06-09

範囲正則化、連合学習で量子化に資する重みの極性クラスタ化

What Do Deepfake Speech Detectors Actually Hear?

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-09

ディープフェイク音声検出器が実際に何を聴くかを可視化

Ethical and Technical Limits of Deepfake Speech Datasets

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-09

ディープフェイク音声39データセットを監査、公平性評価は困難

RAT: Reference-Augmented Training for ASV Anti-Spoofing

arXiv cs.AI (Artificial Intelligence) ・ 2026-06-09

RAT、参照拡張学習でASV成りすまし検知を最先端化

Pushing the Limits of LLM Tool Calling via Experiential Knowledge Integration and Activation

arXiv cs.CL (Computation and Language) ・ 2026-06-09

経験的知識の統合と活性化でLLMのツール呼び出しを強化

ConvMemory v2: A Recall-Preserving Top-10 Evidence Reranker for Conversational Memory Retrieval

arXiv cs.CL (Computation and Language) ・ 2026-06-09

ConvMemory v2、会話記憶検索の想起保持型Top-10再ランカー

Attention-Discounted Adaptive Sampler for Masked Diffusion Language Models

arXiv cs.CL (Computation and Language) ・ 2026-06-09

ADAS、マスク拡散言語モデルの並列復号を再ランクで安全化

K-Forcing: Joint Next-K-Token Decoding via Push-Forward Language Modeling

arXiv cs.CL (Computation and Language) ・ 2026-06-09

K-Forcing、押し出し言語モデルで次kトークンを同時復号

Recovering the Zipfian Distribution in Unsupervised Term Discovery

arXiv cs.CL (Computation and Language) ・ 2026-06-09

教師なし語彙発見、グラフクラスタリングがK-means等を上回る

Continual LLM Upcycling: A Predictor-Gated Bank-Wise Sparsity Training Recipe for Dense-to-Sparse LLMs

arXiv cs.CL (Computation and Language) ・ 2026-06-09

密から疎へ、継続学習でLLMをスパース化するレシピを提示

Attention Expansion: Enhancing Keyphrase Extraction from Long Documents with Attention-Augmented Contextualized Embeddings

arXiv cs.CL (Computation and Language) ・ 2026-06-09

Attention Expansion、長文書のキーフレーズ抽出を強化

REAL: A Reasoning-Enhanced Graph Framework for Long-Term Memory Management of LLMs

arXiv cs.CL (Computation and Language) ・ 2026-06-09

REAL提案、推論強化グラフでLLMの長期記憶を管理

Infini Memory: Maintainable Topic Documents for Long-Term LLM Agent Memory

arXiv cs.CL (Computation and Language) ・ 2026-06-09

Infini Memory、トピック文書でエージェントの長期記憶を保守

Multilingual Word-Level Forced Alignment with Self-Supervised Representations and Learned Dynamic Programming

arXiv cs.CL (Computation and Language) ・ 2026-06-09

自己教師あり表現で多言語の単語強制アラインメントを高精度化

Speaker Group Encoding in Self-supervised Speech Recognition Models

arXiv cs.CL (Computation and Language) ・ 2026-06-09

自己教師あり音声モデルの話者グループ情報の符号化を解析

ParaBridge: Bridging Paralinguistic Perception and Dialogue Behavior in Speech Language Models

arXiv cs.CL (Computation and Language) ・ 2026-06-09

ParaBridge、音声LLMのパラ言語知覚と対話行動の溝を埋める