Ittai Rubinstein
Email: ittai.rubinstein@gmail.com
Hi!
I am a third-year PhD student at MIT advised by Sam Hopkins and supported by the Mathworks EECS Fellowship.
I like all algorithms, but recently my main focus is on data attribution and robust machine learning.
Before coming to MIT, I worked at Qedma, where I led an amazing team of researchers working on quantum error suppresion and mitigation. Before that, I earned my master's degree in computer science from Tel Aviv University, where I was advised by Muli Safra. Before that, I completed my undergraduate studies in mathematics, physics, and computer science at the Technion.
Data attribution models are methods of predicting how a given model would change if samples were added to / removed from its training corpus. Working with Sam Hopkins, my goal has been to answer the question:
Perhaps the most studied type of synchronization channel is the binary deletion channel where each bit of the input is deleted i.i.d. with some fixed probability d. A recent series of works by Con and Shpilka, Tal et al., Pernice et al., and myself ([1], [2], [3], [4]) shows that we can construct efficiently decodable, capacity-achieving error-correcting codes for this channel. This motivates the question of estimating the capacities of such channels.
The best previously known upper bounds on the capacity of this channel were generated by Rahmati and Duman [5], who used the results of running the Blahut-Arimoto Algorithm (BAA - an algorithm for estimating the capacity of synchronous channels) on a series of auxiliary channels with exponentially large alphabets. Running the BAA on these channels is computationally difficult, but we designed a more efficient version of this algorithm (and do a lot of low-level optimization), allowing us to compute better upper bounds on the capacity of the deletion channel.
For the lower bounds, we built on the work of Mitzenmacher and Drinea [6], who show how to convert certain candidate distributions into lower bounds on the capacity of the deletion channel. We adapted the BAA to design better candidate distributions for this construction, resulting in improved lower bounds on the capacity of the binary deletion channel.
Read the full paper on arXiv
Reliable high-accuracy error mitigation for utility-scale quantum circuits
Dorit Aharonov, Ori Alberton, Itai Arad, Yosi Atia, Eyal Bairey, Matan Ben Dov, Asaf Berkovitch, Zvika Brakerski, Itsik Cohen, Eran Fuchs, Omri Golan, Or Golan, Barak D. Gur, Ilya Gurwich, Avieli Haber, Rotem Haber, Dorri Halbertal, Yaron Itkin, Barak A. Katzir, Oded Kenneth, Shlomi Kotler, Roei Levi, Eyal Leviatan, Yotam Y. Lifshitz, Adi Ludmer, Shlomi Matityahu, Ron Aharon Melcer, Adiel Meyer, Omrie Ovdat, Aviad Panahi, Gil Ron, Ittai Rubinstein, Gili Schul, Tali Shnaider, Maor Shutman, Asif Sinay, Tasneem Watad, Assaf Zubida, Netanel H. Lindner
preprint
[arxiv] | [media]
Rescaled Influence Functions: Accurate Data Attribution in High Dimension
Ittai Rubinstein, Sam Hopkins
To appear in NeurIPS 2025
[arxiv] | [github]
Robustness Auditing for Linear Regression: To Singularity and Beyond
Ittai Rubinstein, Sam Hopkins
ICLR 2025
[arxiv] | [github]
The Quasi-Probability Method and Applications for Trace Reconstruction
Ittai Rubinstein
SOSA 2025
[arxiv]
Improved Upper and Lower Bounds on the Capacity of the Binary Deletion Channel
Ittai Rubinstein, Roni Con
ISIT 2023
[arxiv]
Average-Case to (shifted) Worst-Case Reduction for the Trace Reconstruction Problem
Ittai Rubinstein
ICALP 2023
[arxiv]
Explicit and Efficient Construction of (nearly) Optimal Rate Codes for the Binary Deletion Channel and the Poisson Repeat Channel
Ittai Rubinstein
ICALP 2022
[arxiv]
Multivariate Generating Functions for Information Spread on Multi-Type Random Graphs
Yaron Oz, Ittai Rubinstein, Muli Safra
JSTAT 2022
[arxiv]
Heterogeneity and Superspreading Effect on Herd Immunity
Yaron Oz, Ittai Rubinstein, Muli Safra
JSTAT 2021
[arxiv]
Superspreaders and High Variance Infectious Diseases
Yaron Oz, Ittai Rubinstein, Muli Safra
JSTAT 2021
[arxiv]
Deep Learning Reconstruction of Ultrashort Pulses from 2D Spatial Intensity Patterns Recorded by an All-in-Line System in a Single-Shot
Ron Ziv, Alex Dikopoltsev, Tom Zahavy, Ittai Rubinstein, Pavel Sidorenko, Oren Cohen, Mordechai Segev
Optics Express 2020
[arxiv]