Jürgen Schmidhuber

Director, AI Initiative@King Abdullah University of Science and Technology·沙特阿拉伯·OpenReview

研究方向

learning robots · fast weights · curiosity · generative adversarial networks · artificial creativity · unsupervised learning · hierarchical reinforcement learning · deep learning · computer vision · recurrent neural networks · meta-learning · evolutionary computation · reinforcement learning

Jürgen Schmidhuber

Multiple Token Divergence: Measuring and Steering In-Context Computation Density

Huxley-G\"odel Machine: Human-Level Coding Agent Development by an Approximation of the Optimal Self-Improving Machine

Structural Prognostic Event Modeling for Multimodal Cancer Survival Analysis

Efficient Morphology-Control Co-Design via Stackelberg Proximal Policy Optimization

Efficient Multi-Step Reinforcement Learning with Expectation-Maximization Bootstrapping

Decoupling The "What" and "Where" With Polar Coordinate Positional Embedding

All Life is Problem Creation: Learning to Generate Environments that Maximize Performance Gain

Mixture of Sparse Attention: Content-Based Learnable Sparse Attention via MoEs

Know When to Fold 'Em: Predicting an LLM-Judge for Efficient but Performant Inference

Agent-as-a-Judge: Evaluate Agents with Agents

Curious Causality-Seeking Agents Learn Meta Causal World

Fairness Overfitting in Machine Learning: An Information-Theoretic Perspective

Directly Forecasting Belief for Reinforcement Learning with Delays

Measuring In-Context Computation Complexity via Hidden State Prediction

Class-wise Generalization Error: an Information-Theoretic analysis

FACTS: A Factored State-Space Framework for World Modelling

Agent-as-a-Judge: Evaluating Agents with Agents

Scaling Value Iteration Networks to 5000 Layers for Extreme Long-Term Planning

How to Correctly Do Semantic Backpropagation on Language-based Agentic Systems

Scaling Value Iteration Networks to 5000 Layers for Extreme Long-Term Planning

MarDini: Masked Autoregressive Diffusion for Video Generation at Scale