AI People Wiki
Portrait of Ruslan Salakhutdinov
Photo: Steve Jurvetson from Menlo Park, USA, CC BY 2.0. Credit

Builders & open source

Ruslan Salakhutdinov

UPMC Professor of Computer Science, Machine Learning Department, Carnegie Mellon University

Ruslan Salakhutdinov is a UPMC Professor of Computer Science in the Machine Learning Department at Carnegie Mellon University, which Salakhutdinov joined in February 2016 after serving as an assistant professor at the University of Toronto. Salakhutdinov's homepage lists research interests in deep learning, probabilistic graphical models, and large-scale optimization. Salakhutdinov co-authored the 2014 JMLR paper introducing dropout, a training technique that randomly drops units during training to reduce overfitting, and more recently co-authored VisualWebArena, a benchmark for evaluating multimodal agents on realistic visual web tasks. Salakhutdinov is also listed with Meta Superintelligence Labs on a September 2026 paper about agentic meta-reasoning at inference time.12345

At a glance

Current roles
UPMC Professor of Computer Science, Machine Learning Department +1 more
Sector
University
Education
Carnegie Mellon, Toronto
Work listed
2
Honors
5
Sources cited
9
Profile reviewed

Career

Ruslan Salakhutdinov received a PhD in machine learning from the University of Toronto in 2009 and spent two years as a postdoctoral associate at MIT, in the Department of Brain and Cognitive Sciences and the Computer Science and AI Laboratory. From 2011 to 2016 Salakhutdinov was an assistant professor in the Departments of Computer Science and Statistics at the University of Toronto.2

In February 2016 Salakhutdinov joined the Machine Learning Department at Carnegie Mellon University, where Salakhutdinov is now a UPMC Professor of Computer Science. Salakhutdinov is an action editor of the Journal of Machine Learning Research and has served on the senior program committees of conferences including NeurIPS and ICML. A September 2026 paper lists Salakhutdinov with Meta Superintelligence Labs.215

Research

Salakhutdinov's research aims to understand the computational and statistical principles needed to discover structure in large amounts of data, spanning deep learning, probabilistic graphical models and large-scale optimization. With Geoffrey Hinton, Salakhutdinov wrote "Deep Boltzmann Machines" (AISTATS 2009), and Salakhutdinov co-authored the 2014 JMLR paper on dropout.2163

Later work includes "Show, Attend and Tell" (2015) on image captioning with visual attention, and the language models Transformer-XL and XLNet (both 2019), written with Zihang Dai, Zhilin Yang, Quoc Le and others. More recently Salakhutdinov co-authored VisualWebArena (2024), a benchmark for multimodal web agents, and "Thinking Before Thinking" (2026), on scaling agentic inference through meta-reasoning.78945

Work and contributions

  • Dropout: A Simple Way to Prevent Neural Networks from Overfitting Published research

    Co-author, with Nitish Srivastava, Geoffrey Hinton, Alex Krizhevsky, and Ilya Sutskever, of the 2014 Journal of Machine Learning Research paper describing dropout.3

  • VisualWebArena Published research

    Co-author of the 2024 paper introducing VisualWebArena, a benchmark for evaluating multimodal agents on realistic visual web tasks; the homepage links the paper and its code.41

Timeline

  1. Completed a PhD at the University of Toronto and co-authored "Deep Boltzmann Machines" with Geoffrey Hinton.26
  2. Became an assistant professor of computer science and statistics at the University of Toronto.2
  3. Co-authored the JMLR paper "Dropout: A Simple Way to Prevent Neural Networks from Overfitting".3
  4. Co-authored "Show, Attend and Tell", on neural image caption generation with visual attention.7
  5. Joined the Machine Learning Department at Carnegie Mellon University.2
  6. Co-authored Transformer-XL, a language model with context beyond a fixed length.8
  7. Co-authored XLNet, on generalized autoregressive pretraining for language understanding.9
  8. Co-authored VisualWebArena, a benchmark for multimodal agents on realistic web tasks.4
  9. Co-authored "Thinking Before Thinking", listed with Meta Superintelligence Labs.5

Other documented roles

Roles the sources document that are not confirmed as current.

  • Assistant Professor, Department of Computer Science and Department of Statistics (2011–2016), University of Toronto2

Honors and recognition

  • Alfred P. Sloan Research Fellowship · Alfred P. Sloan Foundation21
  • Microsoft Research Faculty Fellowship · Microsoft Research21
  • Canada Research Chair in Statistical Machine Learning · Canada Research Chairs Program2
  • Pioneers of AI Award · NVIDIA2
  • Senior Fellow · Canadian Institute for Advanced Research (CIFAR)2

Where they work

Suggest a correction or removal

Profiles cover public professional work only, and each statement cites a source. If something is out of date, missing or wrong, tell us and link a source. People can also ask for their profile to be corrected or removed.

References

9 sources · profile reviewed Oct 21, 2026 · data updated Oct 21, 2026

  1. Ruslan Salakhutdinov — homepageRuslan Salakhutdinov · Official page · accessed Oct 21, 2026
  2. Ruslan Salakhutdinov — Brief BiosketchRuslan Salakhutdinov · Official page · accessed Oct 21, 2026
  3. Dropout: A Simple Way to Prevent Neural Networks from Overfitting (JMLR 15(56), 2014)Journal of Machine Learning Research · Paper · published 2014 · accessed Oct 21, 2026
  4. VisualWebArena: Evaluating Multimodal Agents on Realistic Visual Web Tasks (arXiv 2401.13649)arXiv · Paper · published Jan 24, 2024 · accessed Oct 21, 2026
  5. Thinking Before Thinking: Scaling Agentic Inference Through Meta-Reasoning (arXiv 2609.38147)arXiv · Paper · published Sep 29, 2026 · accessed Oct 21, 2026
  6. Deep Boltzmann Machines (AISTATS 2009, PMLR volume 5)Proceedings of Machine Learning Research · Paper · published 2009 · accessed Oct 21, 2026
  7. Show, Attend and Tell: Neural Image Caption Generation with Visual Attention (arXiv 1502.03044)arXiv · Paper · published Feb 10, 2015 · accessed Oct 21, 2026
  8. Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context (arXiv 1901.02860)arXiv · Paper · published Jan 9, 2019 · accessed Oct 21, 2026
  9. XLNet: Generalized Autoregressive Pretraining for Language Understanding (arXiv 1906.08237)arXiv · Paper · published Jun 19, 2019 · accessed Oct 21, 2026