🧑‍🔬 Razvan Pascanu

🏛️ Affiliations

🏢Google DeepMindLondon · United Kingdom2015–2026 · 41 papersEmploymentcurrent🎓Université de MontréalMontreal · Canada2010–2026 · 25 papersEducation until 2014current🔬Mila - Quebec Artificial Intelligence InstituteMontreal · Canada2023–2026 · 4 paperscurrent🎓Imperial College LondonLondon · United Kingdom2020–2026 · 2 paperscurrent🏢GoogleMountain View · United States2015–2025 · 35 papers🏢GoogleLondon · United Kingdom2018–2024 · 3 papersEmployment🎓Stanford UniversityStanford · United States2021–2025 · 2 papers🎓ETH ZurichZurich · Switzerland2023–2024 · 2 papers🎓University of OxfordOxford · United Kingdom2019–2022 · 2 papers🎓Washington State UniversityPullman · United States2021–2022 · 2 papers

From the bylines of their papers (OpenAlex, 2026-10-14), ORCID and Wikidata; institutions on papers are matched by machine. 4 more institutions on single bylines are not listed.

🧭 Fields, topics and concepts

Of its works in our library, with the number of works on each.

📅 Works by year

2020s

2010s

📄 Papers

The most cited first (OpenAlex and our library).

1. Theano: new features and speed improvements (2012 · 1,111 citations)
   This 2012 report describes how Theano, a Python-based compiler that turns symbolic linear-algebra expressions into optimized CPU or GPU code, was extended and made faster.
2. A simple neural network module for relational reasoning (2017 · 632 citations)
   Networks such as CNNs and MLPs often fail on questions that depend on how several entities relate. Symbolic systems handle relations well but are brittle and hard to ground in raw input. The authors propose the Relation Network (RN), a small, fully differentiable module. It takes a set of "objects," applies one…
3. Progress & Compress: A scalable framework for continual learning (2018 · 338 citations)
   Neural networks that learn tasks one after another tend to lose earlier skills, a problem called catastrophic forgetting.
4. How to Construct Deep Recurrent Neural Networks (2013 · 302 citations)
   In feedforward networks, depth means stacking nonlinear layers between input and output. In recurrent networks the idea is ambiguous: an RNN unrolled over time is already deep, yet each single step uses only an affine map followed by an elementwise nonlinearity. Pascanu, Gulcehre, Cho and Bengio examine one step of…
5. Resurrecting Recurrent Neural Networks for Long Sequences (2023 · 118 citations)
   Classic RNNs are cheap at inference on long inputs, but they train slowly and suffer from vanishing or exploding gradients.

All works

27 works of theirs in our library, the most cited first; 5 with a page here, the others link their original until we summarise them.

1. Progressive Neural Networks ↗arXiv:1606.04671 · no summary yetAndrei A. Rusu et al.2016 · 1,295 citations
2. Theano: new features and speed improvementsarXiv:1211.5590Frédéric Bastien, Pascal Lamblin, Razvan Pascanu, et al.2012 · 1,111 citations
3. Identifying and attacking the saddle point problem in high-dimensional non-convex optimization ↗arXiv (Cornell University) · no summary yetYann Dauphin et al.2014 · 818 citations
4. On the Number of Linear Regions of Deep Neural Networks ↗arXiv (Cornell University) · no summary yetGuido Montúfar et al.2014 · 811 citations
5. A simple neural network module for relational reasoningarXiv:1706.01427Adam Santoro, David Raposo, David G. T. Barrett, et al.2017 · 632 citations
6. Interaction Networks for Learning about Objects, Relations and Physics ↗arXiv:1612.00222 · no summary yetPeter W. Battaglia et al.2016 · 581 citations
7. Learning Deep Generative Models of Graphs ↗arXiv (Cornell University) · no summary yetYujia Li et al.2018 · 422 citations
8. Learning to Navigate in Complex Environments ↗arXiv (Cornell University) · no summary yetPiotr Mirowski et al.2016 · 366 citations
9. Progress & Compress: A scalable framework for continual learningarXiv:1805.06370Jonathan Schwarz, Jelena Luketina, Wojciech M. Czarnecki, et al.2018 · 338 citations
10. How to Construct Deep Recurrent Neural NetworksarXiv:1312.6026Razvan Pascanu, Caglar Gulcehre, Kyunghyun Cho, et al.2013 · 302 citations
11. Imagination-Augmented Agents for Deep Reinforcement Learning ↗arXiv (Cornell University) · no summary yetThéophane Weber et al.2017 · 241 citations
12. Meta-Learning with Latent Embedding Optimization ↗arXiv (Cornell University) · no summary yetAndrei A. Rusu et al.2018 · 213 citations
13. Learned-Norm Pooling for Deep Feedforward and Recurrent Neural Networks ↗Lecture notes in computer science · no summary yetÇağlar Gülçehre et al.2014 · 171 citations
14. Relational Deep Reinforcement Learning ↗arXiv (Cornell University) · no summary yetVinícius Zambaldi et al.2018 · 160 citations
15. Deep reinforcement learning with relational inductive biases ↗International Conference on Learning Representations · no summary yetVinícius Zambaldi et al.2018 · 132 citations
16. Natural Neural Networks ↗arXiv (Cornell University) · no summary yetDesjardins, Guillaume et al.2015 · 131 citations
17. Resurrecting Recurrent Neural Networks for Long SequencesarXiv:2303.06349Antonio Orvieto, Samuel L Smith, Albert Gu, et al.2023 · 118 citations
18. Policy Distillation ↗arXiv (Cornell University) · no summary yetAndrei A. Rusu et al.2015 · 108 citations
19. Learning model-based planning from scratch ↗arXiv (Cornell University) · no summary yetRazvan Pascanu et al.2017 · 76 citations
20. Metacontrol for Adaptive Imagination-Based Optimization ↗arXiv (Cornell University) · no summary yetJessica B. Hamrick et al.2017 · 46 citations
21. Relational recurrent neural networks ↗arXiv (Cornell University) · no summary yetAdam Santoro et al.2018 · 43 citations
22. Memory-based Parameter Adaptation ↗arXiv (Cornell University) · no summary yetPablo Sprechmann et al.2018 · 42 citations
23. Pointer Graph Networks ↗arXiv:2006.06380 · no summary yetPetar Veličković et al.2020 · 25 citations
24. RecurrentGemma: Moving Past Transformers for Efficient Open Language Models ↗arXiv (Cornell University) · no summary yetAleksandar Botev et al.2024 · 22 citations
25. Learned-norm pooling for deep neural networks. ↗arXiv (Cornell University) · no summary yetÇağlar Gülçehre et al.2013 · 6 citations

See all 27 works by Razvan Pascanu →

🤝 Co-authors here

12 of 19 co-authors with a page, the most papers together first.

• Caglar Gulcehre ×2
• Yoshua Bengio ×2
• Adam Santoro
• Albert Gu
• Antonio Orvieto
• David G. T. Barrett
• David Raposo
• David Warde-Farley
• Frédéric Bastien
• Ian Goodfellow
• Jelena Luketina
• Jonathan Schwarz

Venues

Where their works here were published, the most works first, with the editions of those that have a page.

• International Conference on Learning Representations
• Lecture notes in computer science

🔗 Cited by

Papers of our library cite their works 1,344 times (of the 149,439 papers whose bibliographies we read); 153 citing papers with a page here →.

Updated 2 November 2026

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/authors/razvan-pascanu/