Jarosław Błasiok

Jarosław Błasiok

Postdoctoral Researcher in Computer Science - Simons Junior Fellow

ETH Zürich

Biography

Jarosław Błasiok is a Postdoctoral Researcher at ETH Zürich in the group of Professor David Steurer. Earlier he was a Junior Fellow at Simons Society of Fellows, conducting postdoctoral research at the Theory of Computation group at Columbia University, under mentorship of Professor Alex Andoni. He finished his Ph.D. at the John A. Paulson School of Engineering and Applied Sciences at Harvard University, advised by Professor Jelani Nelson. He received his B.S. and M.Sc. in Computer Science from the University of Warsaw.

He has a broad research interest in Theoretical Computer Science and has worked in design and analysis of streaming algorithms, the theory of error-correcting codes, algorithms related to machine learning, differential privacy and compressed sensing. In his dissertation, he described his research in streaming algorithms and error correction, featuring applications of high dimensional probability in these two areas.

In the recent series of project his research focused on mathematical aspects of tne notion of calibration for a machine learning predictor — which is a quality of a predictor to assess its own uncertainty.

Download my resumé .

Interests
  • Streaming Algorithms
  • Error-correcting codes
  • Communication Complexity
  • High-dimensional probability
Education
  • PhD in Computer Science, 2019

    Harvard University

  • MSc in Computer Science, 2014

    University of Warsaw

  • BSc in Computer Science, 2011

    University of Warsaw

Recent Publications

(2023). Smooth ECE: Principled Reliability Diagrams via Kernel Smoothing. Unpublished manuscript.

Cite arxiv

(2023). When Does Optimizing a Proper Loss Yield Calibration?. NeurIPS 2023.

Cite arxiv

(2023). Matrix Multiplication and Number On the Forehead Communication. CCC 2023.

Cite arxiv

(2023). Loss minimization yields multicalibration for large neural networks. Unpublished Manuscript.

Cite arxiv

(2023). A Unified Theory of Distance from Calibration. STOC 2023.

Cite arxiv

Contact