MLinfo | 機械学習・AI論文まとめ

Training-Free Task Vectors for LLM Behavioral Control

Task vectors enable post-training model editing by identifying semantically meaningful directions in weight sp

MI向き自然言語処理大規模言語モデルテキスト

用途: 技術検証・論文読解補助
難易度: Hard
コスト: High

Adaptive Anisotropic Attention for Axis-Structured Signals

Dense self-attention treats all token pairs as equally plausible before learning, an interaction-isotropic pri

深層学習Transformer音声

用途: 技術検証・論文読解補助
難易度: Hard
コスト: Medium

Inclusive electron-nucleus cross section models from domain adaptation

We apply transfer learning (TL) to construct data-driven models of inclusive electron-nucleus cross sections.

深層学習Transformer

用途: 技術検証・論文読解補助
難易度: Hard
コスト: High

品質予測/異常検知自然言語処理大規模言語モデルテキスト強化学習

Environments as Scaffold: Enriching Feedback to Bootstrap Self-Evolving Agents in Long-Horizon Tasks

Large Language Models demonstrate remarkable proficiency in static reasoning, yet training them as autonomous

用途: 技術検証・論文読解補助
難易度: Hard
コスト: High

Miles v0.1: Production-Level Post-Training

We present Miles v0.1, a full-stack, production-ready system for frontier post-training. Building upon the cle

用途: 技術検証・論文読解補助
難易度: Hard
コスト: High

Geometry-Aware Bayesian Parameter-Efficient Fine-Tuning on the Stiefel Manifold via Stein Variational Gradient Descent

Several geometry-aware approaches to low-rank adaptation have emerged for parameter-efficient fine-tuning of l

用途: 技術検証・論文読解補助
難易度: Hard
コスト: Medium

EMBLEM: Enhancing Multi-script Table Detection through Masking

Table detection is a core task in document analysis, supporting downstream applications such as information re

自然言語処理ファインチューニング検出QA検索

用途: 検出
難易度: Hard
コスト: High

Marigold V2: Revisiting Diffusion Transformers for Monocular Depth Estimation

Monocular depth estimation is a ubiquitous yet highly ill-posed computer vision task, with downstream applicat

深層学習Transformer生成回帰異常検知

用途: 生成
難易度: Hard
コスト: High

Risk-Conditioned Fine-Tuning of Large Language Models

Large Language Models (LLMs) are increasingly deployed in settings where rare but severe harmful generations c

自然言語処理大規模言語モデル生成テキスト

用途: 生成
難易度: Hard
コスト: High

Co-Evolving Harnesses and Models: On-Policy Correction Helps Weaker Models Catch Up Where Imitation Fails

Agent harnesses (the system prompt, tool set, execution hooks, and context-management scaffolding around a mod

深層学習軽量化・量子化テキスト

用途: 技術検証・論文読解補助
難易度: Hard
コスト: High

品質予測/異常検知自然言語処理ファインチューニング

Good Pretraining, Bad SFT: Checkpoint Quality Across the Training Stack

Language-model checkpoints are commonly selected by pretraining loss or benchmark scores, assuming that the hi

用途: 技術検証・論文読解補助
難易度: Hard
コスト: High

PlannerForge: LLM Agents for Scenario-Based Testing of Motion Planners in Autonomous Driving

Ensuring the safety of autonomous driving is a critical challenge. Scenario-based testing is a systematic proc

深層学習Transformer生成テキスト3D

用途: 生成
難易度: Hard
コスト: High

Same Values, Different Languages? From Multilingual Probing to Steering LLMs Toward Chinese Social Values

As Large Language Models (LLMs) are increasingly integrated into human society, aligning them with pluralistic

表形式向き自然言語処理大規模言語モデルテキスト

用途: 技術検証・論文読解補助
難易度: Hard
コスト: High

Feyospace-v1: How the Cyber Mercury Seven Trained Frontier Cyber Models

Training capable cyber agents is often treated primarily as a problem of model scale, yet open-weight post-tra

自然言語処理RAGテキスト

用途: 技術検証・論文読解補助
難易度: Hard
コスト: High

FPicker: Topology-Guided Evolution for Filament Tracing in Low-SNR Microscopy

Automating filament tracing in Cryo-Electron Microscopy (Cryo-EM) is essential for 3D helical reconstruction b

深層学習Transformer検出画像テキスト

用途: 検出
難易度: Hard
コスト: High

MI向き自然言語処理ファインチューニングテキストマルチモーダル

CircuTutor: Transforming Static Circuit Problems into Intelligent and Dynamic Tutoring

Learning direct current circuit concepts requires learners to connect invisible physical quantities, such as c

用途: 技術検証・論文読解補助
難易度: Hard
コスト: High

Bridging the Semantic-Utility Gap in Multimodal RAG via Generator-in-the-Loop Alignment

Vision-language models (VLMs) augmented with retrieval-augmented generation (RAG) benefit from access to exter

自然言語処理RAG生成画像テキスト

用途: 生成
難易度: Hard
コスト: High

OntologyBench: Can Dense Retrieval Satisfy Structured Biomedical Constraints?

We introduce OntologyBench, a tiered biomedical retrieval benchmark comprising 471,854 training and 125,744 ev

自然言語処理大規模言語モデル検索テキスト

用途: 検索
難易度: Hard
コスト: High

Router Prior Bias: Preserving Base Routing Structure in MoE Post-Training

Mixture-of-Experts (MoE) pretraining relies on an auxiliary load-balancing loss (LBL) to drive per-expert util

自然言語処理ファインチューニングテキスト

用途: 技術検証・論文読解補助
難易度: Hard
コスト: High

Inference-Time Nash Alignment

Preference-based fine-tuning methods such as RLHF and DPO require substantial compute and large preference dat

用途: 技術検証・論文読解補助
難易度: Hard
コスト: Medium

品質予測/異常検知自然言語処理大規模言語モデル生成

ActReview: Rebuttal-Guided Training Data and Rubric Rewards for Actionable Peer Review Generation

As LLMs are increasingly used for pre-submission self-review, there is growing demand for feedback that not on

用途: 生成
難易度: Hard
コスト: High

Which Forms of Caregiver Feedback Support Grammar Learning? A Reinforcement-Learning Study of Child-Like Language Models

Social interaction is central to children's language learning, but the effects of different forms of caregiver

自然言語処理大規模言語モデル生成テキスト強化学習

用途: 生成
難易度: Hard
コスト: High

NeoHorse-1: Towards Recursive Self-Improvement via Agentic Post-Training with Routing Harness

Recursive self-improvement (RSI) requires a concrete mechanism through which an AI system observes its capabil

深層学習軽量化・量子化テキスト

用途: 技術検証・論文読解補助
難易度: Hard
コスト: High

Popular Knowledge Propagates More Errors in LLM Knowledge Updating

Updating a language model's knowledge through fine-tuning is essential for keeping its outputs current, yet ca

深層学習軽量化・量子化テキスト

用途: 技術検証・論文読解補助
難易度: Hard
コスト: High

CVT-GS: Learning to Simplify 3D Gaussian Splatting with Centroidal Voronoi Tessellation

While 3D Gaussian Splatting (3DGS) has emerged as a powerful representation for real-time novel view synthesis

深層学習軽量化・量子化生成画像3D

用途: 生成
難易度: Hard
コスト: High

深層学習TransformerQAテキストマルチモーダル

STSG-VQA: Evidence-Grounded Temporal Question Answering from Surgical Spatio-Temporal Scene Graphs

Despite recent advances in surgical vision-language models (VLMs), temporal reasoning remains limited because

用途: QA
難易度: Hard
コスト: High

SoftRerank: Hierarchical Soft Fusion with Candidate-Label Reranking for Long-Tailed Micro-Action Recognition

Micro-actions are subtle, low-intensity non-verbal behaviors that provide cues to fine-grained human states, i

深層学習軽量化・量子化分類画像動画

用途: 分類
難易度: Hard
コスト: High

自然言語処理ファインチューニングセグメンテーション画像3D

MRI-Guided Reslice-Refined Cross-Slice SDF Reconstruction of the Left Ventricle from Cardiac MRI with Sparse Axial Supervision

Reconstructing a three-dimensional left-ventricular (LV) endocardial surface from cardiac magnetic resonance (

用途: セグメンテーション
難易度: Hard
コスト: High

Proxy Policy Steering

Generalist robot policies carry broad manipulation priors from large-scale data, but specializing them to a ne

用途: 技術検証・論文読解補助
難易度: Hard
コスト: High

OmniNav: Robust Long-Horizon Target Navigation in Dynamic Environments

Long-horizon target navigation requires a robot to sustain task execution across evolving observations, decisi

自然言語処理ファインチューニングテキスト3D

用途: 技術検証・論文読解補助
難易度: Hard
コスト: High

MI-PEFT: Mixture-of-Experts Integrated Parameter-Efficient Fine-Tuning Protein Language Models Improves Acidophilic Proteins Classification

Acidophilic proteins that remain stable and functional under highly acidic conditions, are important for indus

深層学習軽量化・量子化分類テキスト

用途: 分類
難易度: Hard
コスト: High

Two-Scale Localized PCA-Net: Coarse-Global and Local-Residual Representations for Artifact-Reduced PDE Operator Learning

Localized dimensionality reduction improves the scalability of operator learning for high-dimensional partial

用途: 技術検証・論文読解補助
難易度: Hard
コスト: Medium

自然言語処理ファインチューニング生成画像マルチモーダル

Foundation Models for Generalizable Semantic and Goal-Oriented Communication

Semantic and goal-oriented communication is increasingly studied for 6G, but generalization beyond seen data r

用途: 生成
難易度: Hard
コスト: High

Sub-6 GHz Over-the-Air AMC via Curriculum Fine-Tuned CNN-Transformers

Automatic modulation classification (AMC) models are frequently trained and validated on synthetic or channel-

深層学習Transformer分類

用途: 分類
難易度: Hard
コスト: Low

MpSub: A Momentum $p$-Dimensional Subspace Trust-Region Method for Derivative-Free Fine-Tuning of Large Language Models

Full-parameter fine-tuning of large language models has substantial memory costs because backpropagation store

自然言語処理大規模言語モデルテキスト

用途: 技術検証・論文読解補助
難易度: Hard
コスト: High

CoRL: Co-Evolutionary Reinforcement Learning for Adaptive Indirect Prompt-Injection Attacks and Defenses

Tool-augmented language agents are vulnerable to indirect prompt injection (IPI). Unlike direct prompt injecti

自然言語処理ファインチューニング強化学習

用途: 技術検証・論文読解補助
難易度: Hard
コスト: Medium

DeepTable: Structural Attention Biases and Tree Path Encoding for Hierarchical Table Understanding

Large language models (LLMs) have demonstrated strong performance in table understanding. However, they typica

深層学習軽量化・量子化QAテキスト

用途: QA
難易度: Hard
コスト: High

Beyond Single-Negative Preference: Multi-Negative DPO for LLM-Centric Historical Entity Linking

Large language models (LLMs) have recently shown promise for historical entity linking, but preference optimiz

自然言語処理大規模言語モデル生成テキスト

用途: 生成
難易度: Hard
コスト: High

品質予測/異常検知深層学習軽量化・量子化テキスト強化学習

arxivGitHubあり2026-09-07

Revisiting Complete Reasoning Traces for Post-Training

Large language models (LLMs) are often post-trained on pre-collected reasoning trajectories to improve their r

用途: 技術検証・論文読解補助
難易度: Hard
コスト: High

センサ/時系列深層学習Transformer検出音声自己教師

Comparing Self-Supervised and Domain-Invariant Features for Cross-Domain Voice Phishing Detection

Voice phishing detection faces three critical challenges: real criminal recordings are unavailable due to priv

用途: 検出
難易度: Hard
コスト: High

説明可能自然言語処理ファインチューニング検出画像テキスト

DroneGround: Open-Vocabulary Drone Payload Characterization Using Synthetic Data and Grounded Vision-Language Models

Automated drone surveillance has become increasingly important for public safety, critical infrastructure prot

用途: 検出
難易度: Hard
コスト: High

Zero-Shot 3D Plant Organ Segmentation with SAM3 and Semantic NeRFs

Accurate 3D plant organ segmentation is fundamental to automated phenotyping. Existing approaches rely on anno

自然言語処理RAGセグメンテーション画像テキスト

用途: セグメンテーション
難易度: Hard
コスト: High

Search-to-World: Evaluation of 3D World Delivery from User Request through Web Search

Agentic systems can interpret user requests, search the live web, and use external tools, but their ability to

自然言語処理ファインチューニング画像動画3D

用途: 技術検証・論文読解補助
難易度: Hard
コスト: High

品質予測/異常検知深層学習軽量化・量子化テキストマルチモーダル強化学習

Stable-MM-R1: Anchoring Multimodal Reasoning Dynamics via Entropy-Guided Stratification

While Reinforcement Learning (RL) effectively incentivizes reasoning in Large Language Models, current pipelin

用途: 技術検証・論文読解補助
難易度: Hard
コスト: High

品質予測/異常検知自然言語処理ファインチューニング生成画像テキスト

Single Image to Textured 3D Object Generation in Frequency Domain: From Theory to Pipeline

Single-view 3D reconstruction, also known as image-to-3D, is a persistently challenging task due to the extrem

用途: 生成
難易度: Hard
コスト: High

表形式向き自然言語処理大規模言語モデル分類検出テキスト

arxivGitHubあり2026-09-07

SupGRPO: Enhancing GRPO with Matching-based Online SFT for Text Spotting

Text spotting requires both accurate text recognition and precise spatial localization. Current specialised sp

用途: 分類
難易度: Hard
コスト: High

GIFT: Goal-Injected Fine-Tuning for Efficient Manipulation Policy Adaptation

Compared with relying solely on initial observations and language instructions, predicting goal images with ge

深層学習軽量化・量子化生成画像マルチモーダル

用途: 生成
難易度: Hard
コスト: High

品質予測/異常検知自然言語処理ファインチューニング画像

SPOT: Spatial Perception-Oriented Long-Horizon Humanoid Teleoperation

High-quality demonstration data is becoming a central bottleneck for training general-purpose humanoid robots.

用途: 技術検証・論文読解補助
難易度: Hard
コスト: High

センサ/時系列自然言語処理RAG画像テキストマルチモーダル

Zero-Shot Sim-to-Real Contact-Rich Assembly via Proprioception-Anchored Cross-Modal Pretraining

Contact-rich assembly remains challenging because it requires submillimeter spatial accuracy and reliable inte

用途: 技術検証・論文読解補助
難易度: Hard
コスト: High

MAC-I$^2$: Learned Metrics-Aware Covariance for Robust Visual-Inertial Fusion in Initialization and Calibration

Visual-Inertial (VI) fusion is fundamental to accurate and robust state estimation, where camera and IMU measu

深層学習Transformer画像テキスト

用途: 技術検証・論文読解補助
難易度: Hard
コスト: High

少数データ向き条件最適化自然言語処理ファインチューニング分類テキスト

Train Smarter, Not Harder: Switching Signal-Guided Training in Active Learning

Training strategy, namely whether to retrain from scratch or fine-tune from the previous checkpoint, is an ove

用途: 分類
難易度: Hard
コスト: High

arxivGitHubあり2026-09-06

Event Interaction in Low-Rank Bottlenecks for Temporal Relation Extraction

Temporal relation extraction determines whether an event occurs before, after, or simultaneously with another

深層学習CNN

用途: 技術検証・論文読解補助
難易度: Easy
コスト: Medium

arxivGitHubあり2026-09-06

Data Efficient Sample Selection for In-Context Learning

The In-context learning (ICL) paradigm aids large language models (LLMs) to adapt to new tasks without need fo

説明可能深層学習軽量化・量子化テキスト

用途: 技術検証・論文読解補助
難易度: Hard
コスト: High

SerenAI: State-transition system inspired by text-based world AI models

Although professional workflows leverage large language models widely, the interpretation for auditing unconst

深層学習軽量化・量子化生成テキスト

用途: 生成
難易度: Hard
コスト: High

品質予測/異常検知自然言語処理RAG分類画像テキスト

Visual Search Augmented Chain-of-Thought Reasoning for Attribute Value Extraction from Product Videos

Existing approaches to visual attribute value extraction (AVE) primarily rely on static product images, failin

用途: 分類
難易度: Hard
コスト: High

ContextFlow: In-Context Flow Matching for Robot Manipulation

Although highly effective in vision and language domains, applying in-context learning to robotics remains cha

自然言語処理RAG生成画像テキスト

用途: 生成
難易度: Hard
コスト: High

MemCorr-DP: Counterfactual Correspondence Conditioning for a Diffusion Policy Guided by a Reference

Behavior-cloned visuomotor policies can remain accurate near their training distribution yet fail when object

深層学習Transformer画像3D

用途: 技術検証・論文読解補助
難易度: Hard
コスト: High

LANTERN: A Closed-Loop Benchmark for VLM-Based Cooperative Driving with Temporally Grounded Warnings

We present LANTERN, a closed-loop benchmark for temporally grounded cooperative warnings. LANTERN separates wa

用途: 技術検証・論文読解補助
難易度: Hard
コスト: High

On-the-go Forgetting without Explicit Unlearning via ERASE

Existing unlearning approaches typically rely on post hoc weight adaptation or distillation, leading to duplic

用途: 技術検証・論文読解補助
難易度: Hard
コスト: High

自然言語処理ファインチューニング検出画像マルチモーダル

Where Success Breaks: Failure-Boundary Learning for Robust Vision-Language-Action Models

Vision-language-action (VLA) models adapted through supervised fine-tuning (SFT) inherit a structural asymmetr

用途: 検出
難易度: Hard
コスト: High

GLoRI: Closed-Loop Whole-Body Tracking with Global-Local Reference Interaction for Humanoid Loco-Manipulation

Humanoid loco-manipulation requires accurate whole-body motion tracking in the world frame for physical intera

用途: 技術検証・論文読解補助
難易度: Hard
コスト: Medium

A4A: Cross-Embodiment Transfer of Action-Oriented 4D Affordances from Human Demonstrations

Human demonstrations contain rich manipulation knowledge, but it remains unclear what information can be trans

自然言語処理ファインチューニング動画3D

用途: 技術検証・論文読解補助
難易度: Hard
コスト: High

自然言語処理大規模言語モデル画像テキストマルチモーダル

Time-Aware Assistive Navigation

Can interactive vision-and-language agents learn not just what to say but also \textbf{\textit{when}} to say i

用途: 技術検証・論文読解補助
難易度: Hard
コスト: High

GE-Act 2.0: Pretraining and Scaling a World-Action Model for Robotic Manipulation

World-action models (WAM) predict future states to guide robot actions, enabling learning from both action-fre

自然言語処理RAG生成異常検知画像

用途: 生成
難易度: Hard
コスト: High

Towards Neuro-Symbolic Procedural Reasoning for Long-Horizon Vision-Language-Action Manipulation

Vision-language-action (VLA) models can execute short manipulation skills, but remain brittle in long-horizon

自然言語処理ファインチューニング画像テキスト動画

用途: 技術検証・論文読解補助
難易度: Hard
コスト: High

Information-Guided Safe Reinforcement Learning for Autonomous Gas Source Localization using sUAS

The autonomous localization of fugitive gas emissions using small Unmanned Aircraft Systems (sUAS) constitutes

自然言語処理ファインチューニング検出3D強化学習

用途: 検出
難易度: Hard
コスト: High

arxivPaper only2026-09-03

AquaBEV: Monocular Underwater BEV Occupancy with 3D Sonar Supervision

Autonomous underwater robots are widely used for exploration, monitoring, and inspection, where safe navigatio

品質予測/異常検知自然言語処理ファインチューニング画像3D

用途: 技術検証・論文読解補助
難易度: Hard
コスト: High

arxivPaper only2026-09-02

Causal Foundation Models

Causal inference is the practice of estimating the effect of a treatment or intervention from data. It traditi

自然言語処理RAGテキスト

用途: 技術検証・論文読解補助
難易度: Hard
コスト: High

arxivGitHubあり2026-09-02

TRACE: Spatiotemporal Contact Memory Graph Network Simulator for Granular Dynamics

Learned graph simulators provide an efficient alternative to high-fidelity solvers for granular dynamics. Howe

深層学習RNN / LSTM3D

用途: 技術検証・論文読解補助
難易度: Hard
コスト: High

arxivPaper only2026-09-01

Measuring consistency via ensemble margin and local prediction variability: Auditing decision systems in the presence of predictive multiplicity

The Rashomon effect is a machine learning phenomenon where equally accurate models produce different predictio

表形式向き深層学習Transformer分類テキスト表形式

用途: 分類
難易度: Hard
コスト: High

arxivPaper only2026-09-01

CircuitsDNA: Discovering Unconventional Multi-Accuracy Arithmetic Circuits via Evolutionary Synthesis

Emerging edge AI workloads increasingly require arithmetic units that can trade computational accuracy for eff

深層学習Transformer生成

用途: 生成
難易度: Hard
コスト: Medium

arxivPaper only2026-08-29

Content Exploration Beyond the Feed: Creator Supply and the Shared Corpus

Industrial recommenders give new content initial views through budgeted exploration, then use early performanc

自然言語処理ファインチューニングテキスト動画

用途: 技術検証・論文読解補助
難易度: Hard
コスト: High

arxivPaper only2026-08-28

Fine-Tuning Autobidders with Group Relative Policy Optimization

Automated bidding (autobidding) is a core component of modern online advertising systems. Within this componen

自然言語処理大規模言語モデル生成テキスト強化学習

用途: 生成
難易度: Hard
コスト: High

arxivPaper only2026-08-27

Beyond Search-Imitation: Prior-Directed Exploration for Searchless Chess

Searchless chess networks reach human master strength from a single forward pass by imitating a stronger teach

自然言語処理ファインチューニング強化学習

用途: 技術検証・論文読解補助
難易度: Hard
コスト: Medium

arxivPaper only2026-08-27

GRAS: Guided Reduced-Variance Proposals and Adaptive Selection for Training-Free Reward Alignment in Discrete Diffusion

Discrete diffusion models have become a strong, widely adopted class of generators for sequence data, and stee

用途: 技術検証・論文読解補助
難易度: Hard
コスト: High

arxivPaper only2026-08-26

Fine-Tuning Whisper for Automatic Speech Recognition in Baniwa: A Preliminary Study

Automatic Speech Recognition (ASR) technologies have achieved remarkable performance in recent years through t

自然言語処理ファインチューニング分類テキスト音声

用途: 分類
難易度: Hard
コスト: Low

条件最適化自然言語処理ファインチューニング生成回帰テキスト

arxivGitHubあり2026-08-24

JANUS: Online Jacobian-Aligned Infill for Black-Box Optimization

Population optimizers such as CMA-ES, DE, and multi-objective evolutionary algorithms drive search mainly thro

用途: 生成
難易度: Hard
コスト: High

arxivPaper only2026-08-21

Event-triggered Implicit Perturbation for Zeroth-Order Fine-Tuning of Spiking Transformers

Zeroth-order (ZO) optimization estimates gradients using only forward-pass evaluations, making it suitable for

深層学習Transformer生成テキスト

用途: 生成
難易度: Hard
コスト: Medium

arxivPaper only2026-08-07

Adaptive Hybrid Particle Swarm Optimization with Gradient Descent

Gradient injection helps Particle Swarm Optimization (PSO) only when the swarm has identified a basin with smo

用途: 技術検証・論文読解補助
難易度: Hard
コスト: Medium

arxivPaper only2026-08-04

Omega-S: A Functional Resilience Index for LLM Fine-Tuning

LLMが外部アクションを取り続けている場合、エージェントのメモリーが古くなったり誤った情報を持ったりする可能性があります。この問題を解決するために、この研究ではSafeCommitという技術を提案しています。SafeCo

深層学習正規化・最適化手法テキスト

用途: LLMによるメモリー・グランド・エージェントの安全なアクション
難易度: Hard
コスト: High

arxivPaper only2026-08-03

Divisive Normalization Shapes Low-Rank Slow Manifolds for Continuous Working Memory

この研究では、記憶を維持し更新する能力を強化するため、繰り返し神経ネットワークを用いて記憶の特性を分析しました。記憶を維持するための神経計算の一種として、ダイショビュールノーマリゼーション（Divisive Normal

深層学習Transformer

用途: 瞬時記憶の能力を強化する
難易度: Hard
コスト: High

arxivPaper only2026-07-21

Towards chemistries in dynamical systems

ディナミカルシステムを化学的なパラフレーズで説明する手法を提案し、システムの

用途: ダイナミカルシステムの化学的説明
難易度: Hard
コスト: Medium