Eran Treister
Welcome to my website!
I am a faculty member at the Computer Science Department at Ben Gurion University of the Negev, Beer Sheva, Israel. Before that I completed my post doctoral fellowship at the Department of Earth, Ocean and Atmospheric Sciences in the University of British Columbia, Vancouver (2014-2016). There I worked with Prof. Eldad Haber on geophysical inverse problems.
I obtained my PhD from the Computer Science Department at the Technion (2014), under the supervision of Prof. Irad Yavneh. My dissertation: "Aggregation-based adaptive algebraic multigrid for sparse linear systems."
Research interests:
My broad research field is computational science (or scientific computing). I am interested in developing efficient numerical solution techniques for various large-scale mathematical problems, such as parameter estimation and data-fitting inverse problems, non-linear optimization problems, solutions of partial differential equations, and sparse linear systems. The problems that I consider arise from diverse fields such as machine learning, signal/image processing, and geophysical imaging, for which I develop efficient parallel algorithms and software for solving the problems at hand efficiently, concerning computing time and energy consumption. Such efficient and scalable algorithms and software are particularly important these days, as the available data get larger in various applications and more sophisticated and better algorithms are required for analyzing them.
Current and near-future research projects:
Graph neural networks and graph learning.
Accelerating and understanding deep learning: That includes efficient parametrizations of CNNs, understanding stability issues in DL, efficient prunning and sparsity-based techniques, quantization and compression of CNNs, optimization techniques for CNNs.
Seismic imaging: robust and efficient solution to seismic full waveform inversion (FWI) and optical diffraction tomography (ODT), possibly using deep learning.
Efficient methods for elastic and acoustic wave modelling in frequency domain. Including Deep Learning approaches. Multi-preconditioning.
Graph Neural Networks for accelerating solution of PDEs with spatially adaptive meshes in conjunction with Finite Elements Methods.
3D shape reconstruction and model regularization using parametric level set methods.
Teaching:
Winter 2022/3: Introduction to Data Science.
Winter 2022/3: Numerical Methods for PDEs.
Spring 2023: Optimization Methods with Applications.
Spring 2023: Mini-project in Scientific Computing.
Winter 2022: Practical Deep Learning.
Contact Information:
email: erant at cs.bgu.ac.il
Dept of Computer Science,
Building 37 Office 206,
Ben Gurion University of the Negev, P.O 653, Beer Sheva, Israel, 8410501.
What's new?
Apr 2023: Accepted to ICML: Moshe Eliasof, Lars Ruthotto, Eran Treister, Improving Graph Neural Networks with Learnable Propagation Operators. (arXiv)
Apr 2023: Accepted to ICML: Moshe Eliasof, Fabrizio Frasca, Beatrice Bevilacqua, Eran Treister, Gal Chechik, Haggai Maron, Graph Positional Encoding via Random Feature Propagation. (arXiv)
Apr 2023: New paper online: Moshe Eliasof, Eldad Haber, Eran Treister, DRIP: Deep Regularizers for Inverse Problems, submitted 2023, (arXiv)
Mar 2023: New paper online: Eran Treister and Rachel Yovel, A Hybrid Shifted Laplacian Multigrid and Domain Decomposition Preconditioner for the Elastic Helmholtz Equations, submitted 2023. (arXiv)
Feb 2023: New paper online: Yakov Medvedevsky, Eran Treister, Tirza Routtenberg, Efficient Graph Laplacian Estimation by a Proximal Newton Approach, submitted 2023. (arXiv)
Jan 2023: New MSc student joins the group: Mr. Roi Mashiah. (Jointly supervised with Tirza Routtenberg).
Jan 2023: Ido Ben-Yair passed his candidacy exam and got his MSc degree.
Jan 2023: Accepted to ICLR: Ashkenazi et. al., NeRN -- Learning Neural Representations for Neural Networks (arXiv).
Jan 2023: Ruth Akiva Hochman has finished and got her MSc degree.
Dec 2022: I got promoted to the rank of Associate Professor.
Dec 2022: Accepted to Sensors special issue on DL on Edge Devices: Benjamin J. Bodner, Gil Ben Shalom, Eran Treister. GradFreeBits: Gradient Free Bit Allocation for Mixed Precision Neural Networks
Dec 2022: 4-page abstract accepted to Learning on Graphs (LoG) conference: Moshe Eliasof, Eran Treister. Global-Local Graph Neural Networks for Node-Classification
Sep 2022: Accepted to SISC: Moshe Eliasof, Jonathan Ephrath, Lars Ruthotto, Eran Treister. MGIC: Multigrid-in-Channels Neural Network Architectures (arXiv)
Sep 2022: Accepted to NeurIPS: Shahaf Finder, Yair Zohav, Maor Ashkenasi, Eran Treister, Wavelet Feature Maps Compression for Image-to-Image CNNs. (arXiv)
Aug 2022: Gal Shalom has finished his MSc (Technion), and Ruth Akiva Hochman has submitted her MSc thesis.
Aug 2022: Accepted to ECCV Workshop: Ruth Akiva Hochman, Shahaf E. Finder, Javier S. Turek, Eran Treister, Searching for N:M Fine-Grained Sparsity of Weights and Activations in Neural Networks. ECCV Workshop on Computational Aspects of Deep Learning (CADL), Oct 2022.
Aug 2022: Accepted to Springer RMS: Ido Ben-Yair, Gil Ben Shalom, Moshe Eliasof, Eran Treister, Quantized convolutional neural networks through the lens of partial differential equations, 2022. (arXiv)