AI Wiki — Knowledge Base

【AIの難問は、生命の難問へ：英国王立協会が紐解く「世界モデル」と知能の未来】近年のAIの急速な発展により、AGI、すなわち人間並みの知能はすでに実現したという意見も聞かれるようになりました。果たしてそうなのでしょうか。鍵となるのが「世界モデル（World Model）」の概…

【AIの難問は、生命の難問へ：英国王立協会が紐解く「世界モデル」と知能の未来】近年のAIの急速な発展により、AGI、すなわち人間並みの知能はすでに実現したという意見も聞かれるようになりました。果たしてそうなのでしょうか。鍵となるのが「世界モデル（World Model）」の概念です。世界モデルとは、生き物やAIが外の世界を内部に写し取り、次に何が起きるかを予測して行動するための土台となるもの

researchdavid-ha--x

13 Sept 2026

Search: “research”

ProClosure: Hierarchical Room-Object Assignment using Progressive Boundary Closure from Monocular Video

Quantifying Consonant Contributions to Word Intelligibility via Acoustic Masking

Quantifying the Value of Privileged Information Using a PAC-Bayesian Approach

Quantile-based Loss Filtering for Outlier-Robust Stochastic Gradient Descent

R2VC: Modular Fact-Checking with Retrieval, Verification, and Confidence Calibration

Receiver-Surface Hit Patterns via Legendre Approximation for Molecular Signal Detection

Reinforcement Learning for Syndrome Extraction

Revisiting Multi-Object Tracking Baselines: Hyperparameter Optimization with Multi-Fidelity Greedy Coordinate Search

RoES: Rotational Equivariant Selective-frequency Fusion for Multimodal Images

RunningTensor: Generalizing Linear Attention to Higher-Order Recurrent States

Same Encoder, Different Winner: A Paired-View Framework for Cell Painting Encoder Evaluation

Sampling via Decision-Flow: Training-Free Extraction of Improved Latent Reasoning Paths in Large Language Models

Scalable Discrete-to-Continuous Channel Simulation for Compression and Privacy

Score-based Outlier Generation via Controlling the Radon-Nikodym Derivative

SCORE: SubDistribution-aware Collaborative Knowledge Reinforcing for Cloth-Hybrid Lifelong Person Re-Identification

Self-supervised Pre-training Helps Retinal Disease Progression Modelling Most When Data Is Scarce

Self-Verifying Anomaly Detection using Explainable AI for Cybersecurity of DER Networks

SIFPBPNet: A Dual-Path Network for Wearable and Cuffless Blood Pressure Estimation via Individualized Steady-state Representation

Simplifying ROS2 controllers with a modular architecture for robot-agnostic reference generation

SIMS: Scale-Invariant Merit-Function-Based Scalarization for Multi-Task Learning

Single-Query Person-Centric Bimanual Hand-Object Interaction Detection

Spectral Consistency-Guided Multiview Point Cloud Registration for Low-Overlap Scenes

Split Conformal Prediction with Label-Shift-Adjusted Bayesian Scores

Statistical Mechanics of Semantic Compression

SV-Cine: Diagnosis-Conditioned Segmentation of Single Ventricle Physiology via Generative Data Augmentation

Temporal Recurrence Favors Fewer Layers

The AI industry has taken a doomer turn. What now?

The Cost of Compression: A Rate-Distortion Limit on Factual Hallucination

The Iterative Equivariant Filter

The Rank the Task Demands: A Causal Rank Law for Matrix Memories Trained on Group Composition

Tight Sampling Complexity with stochastic gradient oracles in Fixed Dimensions

TokenMapper: A Step Toward Interoperable Speech Token Translation

Towards Sustainable Hydrogen Systems: Supply Chain Optimization with Model Predictive Control and Reinforcement Learning

Tunable Latent Generative Priors for Compressed Sensing and Inverse Problems

Type Diversity Enables Transformers to Generalise Compositionally

Understanding Whole-Body Robot Teleoperation Strategies Under Diverse Task Objectives and Constraints

Unlabeled Echoes: Pseudo-Labels and Genus-Aware Smoothing for Bat Call Recognition

USPLIT-VQA: U-Shaped Split Learning for Visual Question Answering with Contribution-Aware Weighted Aggregation

VertiFuseX: Generalizable Financial Forecasting via Multi-Stream Temporal Fusion

Very Exciting: Zero-Shot Model Predictive Control of Buildings via Excitation-Based Generalized Transfer Learning Models

What an odour descriptor corpus can and cannot measure: valence, attenuation, and the ceiling of the public record

When Context Misleads: Surprisal, Energy and Attention Entropy as Metrics of Coherence Illusions in LLMs

Write on Paper and Get the Online Digital Trace:newline A New Era for Handwriting

I must be among an extremely small group of people (n=1?) that have both 1) trained a frontier LLM and 2) designed and synthesized custom vi…

Investors in Anthropic founded Open Philanthropy, the leading org promoting AI Doom They have a stake in regulation on AI that excludes comp…

Looped transformers are a popular architecture topic right now. This new technical report extends the loop across tokens. Recurrent Looped T…

A machine-checked proof of the Dong-Yang classification of optimal (n,4) binary codes for BSCs