PeopleofSuperintelligence

Researcher

Roi Livni

Roi Livni is a machine learning theorist whose papers connect private learning, online learning and memorization in stochastic convex optimization. A PhD student at the Hebrew University of Jerusalem from 2008 to 2017, they coauthored "Vanishing Component Analysis", which shared the ICML 2013 best paper award, and a 2014 study of the computational complexity of training neural networks. After a postdoc at Princeton University in 2017 and 2018, they joined Tel Aviv University in 2018. Their 2020 paper with Mark Bun and Shay Moran on private classification and online prediction received a FOCS Best Paper Award, a 2021 paper with Steve Hanneke and Moran was a COLT best paper runner-up, and their 2024 paper on memorization in stochastic convex optimization was an ICML Best Paper. At Tel Aviv University, they are an associate professor in the School of Electrical Engineering.

Timeline

  1. 2008–2017

    PhD at Hebrew University

    From 2008 to 2017, they are a PhD student at the Hebrew University of Jerusalem.

  2. 2013

    A shared ICML best paper

    In 2013, they coauthor "Vanishing Component Analysis" with David Lehavi, Sagi Schein, Hila Nachliely, Shai Shalev-Shwartz and Amir Globerson; it describes a numerically stable procedure that constructs generators of a vanishing ideal. ICML 2013 shares its best paper award between this paper and a paper on submodular function optimization.

  3. 2014

    Complexity of training neural networks

    In 2014, they post "On the Computational Efficiency of Training Neural Networks" on arXiv with Shai Shalev-Shwartz and Ohad Shamir. The paper revisits the computational complexity of training neural networks and gives positive and negative results, some of which yield provably efficient algorithms for training certain types of networks.

  4. 2017–2018

    From Princeton to Tel Aviv

    A postdoc at Princeton University from 2017 to 2018, they teach the course COS 511, Theoretical Machine Learning, in 2017. In 2018, they join Tel Aviv University as an assistant professor.

  5. 2020

    FOCS best paper on private learning

    In 2020, as a faculty member in the Department of Electrical Engineering at Tel Aviv University, they prove with Mark Bun and Shay Moran that every concept class with finite Littlestone dimension can be learned by an approximate differentially private algorithm. The paper, "An Equivalence Between Private Classification and Online Prediction", receives a FOCS 2020 Best Paper Award; together with a converse result by Alon and colleagues, it gives an equivalence between online learnability and private PAC learnability.

  6. 2021

    A COLT best paper runner-up

    In 2021, COLT names "Online Learning with Simple Predictors and a Combinatorial Characterization of Minimax in 0/1 Games", by Steve Hanneke, Roi Livni and Shay Moran, one of its three best paper runners-up.

  7. 2024

    An ICML 2024 Best Paper

    In 2024, they coauthor "Information Complexity of Stochastic Convex Optimization: Applications to Generalization, Memorization, and Tracing" with Idan Attias, Gintare Karolina Dziugaite, Mahdi Haghifam and Daniel Roy, and ICML 2024 names it a Best Paper. The paper characterizes the tradeoff between the accuracy of a learning algorithm and the information it reveals about its training data. Also in 2024, they coauthor "Credit Attribution and Stable Compression" with Shay Moran, Kobbi Nissim and Chirag Pabbaraju, which proposes relaxations of differential privacy for credit attribution by machine learning algorithms.

  8. Since 2024

    Associate professor at Tel Aviv

    In 2024, they become an associate professor in the School of Electrical Engineering at Tel Aviv University.

Career

  1. Since 2018Tel Aviv UniversityAssistant Professor, Associate Professor
  2. 2017–2018Princeton UniversityPostdoctoral Researcher

Education

  1. 2017Hebrew University of JerusalemPhD

From the biography.

  1. The Massachusetts Institute of Technology trained or employed 44 of the 187 AI major-prize winners, more than any other universityIncludes Roi Livni. The Massachusetts Institute of Technology trained or employed 44 of the 187 AI major-prize winners, more than any other university. We compare doctorate and 5+ year employment links, career timing and current workplaces for 187 major-prize and 1,111 best-paper winners.
  2. More people shaping AI studied at Massachusetts Institute of Technology than at any other institutionIncludes Tel Aviv University. We estimate where the people shaping AI studied and where their careers led. Massachusetts Institute of Technology leads with 467 former students, including 291 recorded doctorates.
  3. The company share of AI researchers’ top papers was 31% in 2020–2024 and 8% in 1970–1974Includes Tel Aviv University. We estimate where the people shaping AI worked when their most-cited papers appeared: the company share was 31% in 2020–2024 and 8% in 1970–1974, and Google led all organizations in 2020–2024.
  4. Google has the highest combined h-index among AI organizationsIncludes Tel Aviv University. We estimate that Google holds the highest combined h-index at 8,016, while universities account for 73% of the deduplicated total across AI organizations.

Colleagues

People named in this biography, and people with dated positions at the same organization in the same years.

Sources and credits

Photo: We have no published photo of Roi Livni yet. Send us a free-licensed photo.

  1. Roi Livni — Lacuna lacuna.tiptreesystems.com. Checked 22 October 2026.
  2. Awards | ICML Atlanta (2013) icml.cc. Checked 22 October 2026.
  3. Vanishing Component Analysis (PMLR v28) proceedings.mlr.press. Checked 22 October 2026.
  4. On the Computational Efficiency of Training Neural Networks (arXiv:1410.1141) arxiv.org. Checked 22 October 2026.
  5. COS 511 Theoretical Machine Learning: Home Page (Princeton, Spring 2017) cs.princeton.edu. Checked 22 October 2026.
  6. FOCS 2020 Papers ieee-focs.org. Checked 22 October 2026.
  7. An Equivalence Between Private Classification and Online Prediction (arXiv:2003.00563) arxiv.org. Checked 22 October 2026.
  8. Shay Moran, Learning Theory and Theoretical Computer Science (Technion) shaymrn.cs.technion.ac.il. Checked 22 October 2026.
  9. COLT 2021 - Paper Awards learningtheory.org. Checked 22 October 2026.
  10. ICML 2024 Awards icml.cc. Checked 22 October 2026.
  11. Credit Attribution and Stable Compression (arXiv:2406.15916) arxiv.org. Checked 22 October 2026.
  12. Roi Livni, Tel Aviv University (Israeli universities research portal) cris.iucc.ac.il. Checked 22 October 2026.