Leo Schwinn

Sampling-aware Adversarial Attacks Against Large Language Models

Model Collapse Is Not a Bug but a Feature in Machine Unlearning for LLMs

A Generative Approach to LLM Harmfulness Mitigation with Red Flag Tokens

Effective Data Pruning through Score Extrapolation

Byte Pair Encoding for Efficient Time Series Forecasting

A Probabilistic Perspective on Unlearning and Alignment for Large Language Models

FOCUS: Internal MLLM Representations for Efficient Fine-Grained Visual Question Answering

Flow Matching with Gaussian Process Priors for Probabilistic Time Series Forecasting

Joint Relational Database Generation via Graph-Conditional Diffusion Models

When to retrain a machine learning model

Adversarial Robustness of Graph Transformers

When to retrain a machine learning model

Efficient Time Series Processing for Transformers and State-Space Models through Token Merging

Efficient Time Series Processing for Transformers and State-Space Models through Token Merging

A Unified Approach Towards Active Learning and Out-of-Distribution Detection