Bernhard Schölkopf

Skill Learning via Policy Diversity Yields Identifiable Representations for Reinforcement Learning

Generation is Required for Data-Efficient Perception

Learning Nonlinear Causal Reductions to Explain Reinforcement Learning Policies

CauSciBench: A Comprehensive Benchmark on End-to-End Causal Inference for Scientific Research

Proper Velocity Neural Networks

Riemannian Networks over Full-Rank Correlation Matrices

When Ethics and Payoffs Diverge: LLM Agents in Morally Charged Social Dilemmas

OVid: Open Large-Scale Video Dataset as a Novel Source for Image-Text Data

Physics of Learning: A Lagrangian perspective to different learning paradigms

Direct Advantage Estimation for Scalable and Sample-efficient Deep Reinforcement Learning

Building Transformation Layers for Riemannian Neural Networks

Scaling Behavior of Discrete Diffusion Language Models

On the Limits of Curriculum Learning for Post-Training Large Language Models

Towards Training One-Step Diffusion Models Without Distillation

Rigidity-Aware Geometric Pretraining for Protein Design and Conformational Ensembles

What Can an LLM Flip if It Fails to Flip Coins?

Beyond Additive Noise: DP for LoRA via Random Projections

Do-PFN: In-Context Learning for Causal Effect Estimation

Can Large Language Models Understand Symbolic Graphics Programs?

Reparameterized LLM Training via Orthogonal Equivalence Transformation

Language Model Alignment in Multilingual Trolley Problems

MathGAP: Out-of-Distribution Evaluation on Problems with Arbitrarily Complex Proofs

Standardizing Structural Causal Models

Corrupted by Reasoning: Reasoning Language Models Become Free-Riders in Public Goods Games

SPARTAN: A Sparse Transformer World Model Attending to What Matters

Flipping Against All Odds: Reducing LLM Coin Flip Bias via Verbalized Rejection Sampling

Are Language Models Efficient Reasoners? A Perspective from Logic Programming

Preference Elicitation for Offline Reinforcement Learning

The Directionality of Optimization Trajectories in Neural Networks

Identifiable Exchangeable Mechanisms for Causal Structure and Representation Learning

Counterfactual reasoning: an analysis of in-context emergence

Conformal Generative Modeling with Improved Sample Efficiency through Sequential Greedy Filtering

Direct Advantage Estimation in Partially Observable Environments

Learning Joint Interventional Effects from Single-Variable Interventions in Additive Models

Generative Intervention Models for Causal Perturbation Modeling

Generalized Interpolating Discrete Diffusion

Verbalized Machine Learning: Revisiting Machine Learning with Language Models