Ev Zisselman

Bio

I'm a PhD student at the Electrical & Computer Engineering (ECE) department at the Technion. I'm a member of the Robot Learning Lab, under the guidance of Prof. Aviv Tamar. I have an M.Sc. from the Technion ECE, advised by Prof. Michael Elad. Previously, I obtained a B.Sc. in Electrical & Computer Engineering and a B.Sc. in Physics from the Technion.

Email / Google Scholar / Github / LinkedIn / CV

Research

I am interested in designing machine learning algorithms for reliable decision-making in complex environments. My main focus is generalization in deep reinforcement learning, learning from demonstrations, exploration, and robotics.

News

Our paper "Blindfolded Experts" has been accepted to NeurIPS'25.
Our paper "Blindfolded Experts" has received the Best-paper award at the EXAIT workshop at ICML'25!
Received the PBC’s Fellowships for Postdoc in AI!
Received the Schmidt Postdoc Award for Women in Mathematical and Computing Sciences!

Publications

	Blindfolded Experts Generalize Better: Insights from Robotic Manipulation and Videogames Ev Zisselman, Mirco Mutti, Shelly Francis-Meretzki, Elisei Shafer, Aviv Tamar, NeurIPS, 2025 ICML EXAIT, 2025, (Best paper award) paper / website / talk We propose to clone the behavior of "blindfolded" experts that are compelled to employ non-trivial exploration to solve the task, which leads to better generalization.
	Explore to Generalize in Zero-Shot RL Ev Zisselman, Itai Lavie, Daniel Soudry, Aviv Tamar, NeurIPS, 2023 paper / code / arXiv ExpGen uses exploration during test time for zero-shot generalization in reinforcement learning.
	Regularization Guarantees Generalization in Bayesian Reinforcement Learning through Algorithmic Stability Aviv Tamar, Daniel Soudry, Ev Zisselman, AAAI, 2022 (Oral) paper / arXiv We develop PAC bounds for Bayesian RL (meta-RL). A key underlying result is showing algorithmic stability for regularized MDPs.
	Deep Residual Flow for Out of Distribution Detection Ev Zisselman, Aviv Tamar, CVPR, 2020 paper / code / arXiv An expressive density model based on normalizing flows for out of distribution detection.
	Local Block Coordinate Descent (LoBCoD) Algorithm for the CSC Model Ev Zisselman, Jeremias Sulam, Michael Elad, CVPR, 2020 paper / code / arXiv The Convolutional Sparse Coding (CSC) Model is usfull for various image processing tasks, and has strong connection to Convolutional Neural Networks (CNN).
	Compressed Learning for Image Classification: A Deep Neural Network Approach Ev Zisselman, Amir Adler, Michael Elad, Handbook of Numerical Analysis, Elsevier, 2018 paper An end-to-end deep learning approach for Compressed Learning (CL).