Pietro Berkes Principal Data Scientis at Nagra, Kudelski Lausanne, Switzerland

Publications - Software

Research projects

My main research interest concerns the way the brain forms a high-level representation of the environment from raw sensory input and without supervision. I approach this problem at a computational level in the framework of Bayesian statistical inference and learning. I like to work from different perspectives, using machine learning models and analysis of electrophysiological data. My goal is to understand how neuronal populations support these probabilistic computations, what are the principles that guide their adaptation to the statistics of the environment, and how their final representation of the world is organized.

8. No evidence for active sparsification in primary visual cortex
7. Evidence for statistically optimal inference and learning in the visual cortex
6. Characterizing neural dependencies with Poisson copula models
5. On sparsity and overcompleteness in image models
4. Structured representations in the visual cortex
3. Relation between Slow Feature Analysis and Independent Component Analysis
2. Analysis and interpretation of inhomogeneous quadratic forms as receptive fields
1. Slowness as a computational principle for the visual cortex

Bonus tracks

2. masterbaboon.com: Artificial Life, Artificial Intelligence, and games
1. Promoting transparency in science (outdated)