Christoph H. Lampert

Professor@ISTA (Institute of Science and Technology Austria)·奥地利·OpenReview

研究方向

differential privacy · robustness · fairness · federated learning · transfer learning · continual learning · meta-learning · domain adaptation · statistical learning theory · PAC Bayes · generalization bounds · Machine Learning

7.5

Back to Square Roots: An Optimal Bound on the Matrix Factorization Error for Multi-Epoch Differentially Private SGD

ICLR 2026Poster

通讯

4.7

Federated Learning with Unlabeled Clients: Personalization Can Happen in Low Dimensions

ICLR 2026Rejected

三作

4.7

ASIDE: Architectural Separation of Instructions and Data in Language Models

ICLR 2026Poster

通讯

4.4

From Low Intrinsic Dimensionality to Non-Vacuous Generalization Bounds in Deep Multi-Task Learning

ICLR 2026Rejected

三作

8.2

Neural Collapse is Globally Optimal in Deep Regularized ResNets and Transformers

合作者 (19)

Christoph H. Lampert

Back to Square Roots: An Optimal Bound on the Matrix Factorization Error for Multi-Epoch Differentially Private SGD

Federated Learning with Unlabeled Clients: Personalization Can Happen in Low Dimensions

ASIDE: Architectural Separation of Instructions and Data in Language Models

From Low Intrinsic Dimensionality to Non-Vacuous Generalization Bounds in Deep Multi-Task Learning

Neural Collapse is Globally Optimal in Deep Regularized ResNets and Transformers

Fast Rate Bounds for Multi-Task and Meta-Learning with Different Sample Sizes

Can LLMs Separate Instructions From Data? And What Do We Even Mean By That?

Differentially Private Federated $k$-Means Clustering with Server-Side Data

Continual Release Moment Estimation with Differential Privacy

Differentially Private Federated $k$-Means with Server-Side Data