Shay Deutsch

I am an Assistant Adjunct Professor in the UCLA Department of Mathematics, working with Professor Andrea Bertozzi. I am also a member of the UCLA Computer Vision Labworking with Professor. Stefano Soatto.  

Contact Info:  
University of California, Los Angeles (UCLA), Department of Mathematics 
Mathematical Sciences Building, room 7370
520 Portola Plaza, Los Angeles, CA 90095. 

Research Interests

My research is primarily related to multi-scale representation and robust statistical estimation applied to developing graph based methods for unsupervised learning. Recently I have been applying my work to zero shot learning and topology estimation of 3D point clouds. I am also very interested in networks, deep learning and data-driven dynamical system methods.


I finished my PhD  in the Computer Science Department at the University of Southern California in 2016. My thesis advisor was Professor Gerard Medioni with whom I was working on developing a global framework for Tensor Voting, and on Manifold Learning. I also collaborated with Professor Antonio Ortega on developing a new framework for Manifold Denoising using tools from Graph Signal Processing.

I hold B.Sc. in Mathematics from the Technion, Israel Institute of Technology and M.Sc. in Applied Mathematics from Tel Aviv University. My Master’s thesis advisors were Professor Amir Averbuch and Professor Shai Dekel. My M.Sc. research focused primarily on an alternative method for simultaneous image acquisition and compression called Adaptive Compressed Sensing. The Adaptive Compressed Sensing framework replaces the universal acquisition of incoherent measurements used in the 'classic Compressed Sensing' framework with a fast and direct method for adaptive wavelet tree acquisition.


S. Deutsch, I. Masi and S. Soatto, "Finding Structure in Point Cloud Data with the Robust Isoperimetric Loss", accepted to SSVM19 (Oral Presentation).

S. Deutsch, A. Bertozzi and S. Soatto,"Zero Shot Learning Using the Isoperimetric Loss", Submitted, [pdf]

S. Deutsch, A.Ortega and G.Medioni, "Robust Denoising of Piece-Wise Smooth Manifolds", ICASSP 2018, [pdf]

S. Deutsch, S. KolouroK. KynugnamY. Owechko, and S. Soatto, “Zero Shot Learning Via Multi-Scale Manifold Regularization”, CVPR 2017, pp. 7112-7119, [pdf]

S. Deutsch, A. Ortega, and G. Medioni. “Graph Manifold Based Frequency Analysis For Denoising",  Submitted.

S. Deutsch and G. Medioni. "Learning the Geometric Structure of Manifolds with Singularities Using the Tensor Voting Graph", Journal of Mathematical Imaging and Vision2017, 402-422.

S. Deutsch, A. Ortega, and G. Medioni. “Manifold Denoising Based on Spectral Graph Wavelets",  ICASSP 2016.

S. Deutsch and G. Medioni. “Intersecting Manifolds: Detection, Segmentation, and Labeling”,  International Joint Conference on Artificial Intelligence,  (IJCAI), 2015. [pdf]

S. Deutsch and G. Medioni. “Unsupervised Learning Using the Tensor Voting Graph”, Scale Space and Variational Methods in Computer Vision (SSVM), 2015. [pdf]

A. Averbuch, S. Dekel and S.Deutsch. “Adaptive Compressed Image Sensing Using Dictionaries”. Siam Journal Of Imaging Sciences, 5(1), (2012), 57-89.[pdf]

S. Deutsch, A. Averbuch and S. Dekel. “Adaptive compressed image sensing based on wavelet modeling and direct sampling.”, Sampling Theory and Applications (SAMPTA) Conference, May 2009. Marseille, France.[pdf]


Spring 2019: MATH 151B - Applied Numerical Methods (II)

Fall 2018: MATH 151B - Applied Numerical Methods (II)

Spring 2018: MATH 151B - Applied Numerical Methods (II)

Winter 2018:MATH 170A - Probability Theory

Spring 2017: MATH 151A - Applied Numerical Methods (I)

Winter 2017:MATH 151A - Applied Numerical Methods (I)