JB
Joan Bruna
cs.LGstat.MLcs.CVmath.OCcs.AIcs.NEmath.PRcs.CLcs.GRcs.NA
On Valency
published · living versionsW_9msk58k4·v1 · currentpublished
Intriguing properties of neural networks
with Christian Szegedy, Wojciech Zaremba, Ilya Sutskever, Dumitru Erhan +2
1 version
Preprints & journals
119 papers in the corpus · 2010–2026Lost in Tokenization: Fundamental Trade-offs in Graph Tokenization for Transformers2605.22471v2 · Maya Bechler-Speicher, Gilad Yehudai, Gil Harari et al.2026 · 0 citationsarXiv
A Functional-Space Mean-Field Theory of Partially-Trained Three-Layer Neural Networks2210.16286v2 · Zhengdao Chen, Eric Vanden-Eijnden, Joan Bruna2022 · 0 citationsJournal of Machine Learning Research, 27(52):1-67, 2026
The Multiscale Single-Index Model: A Stylized Model for Hierarchical Feature Learning2607.03347v1 · Joan Bruna2026 · 0 citationsarXiv
Learning Stable Group Invariant Representations with Convolutional Networks1301.3537v1 · Joan Bruna, Arthur Szlam, Yann LeCun2013 · 30 citationsarXiv
Uniform-in-Time Weak Propagation-of-Chaos in Shallow Neural Networks2605.22010v1 · Margalit Glasgow, Joan Bruna2026 · 0 citationsarXiv
Generative Modeling from Black-box Corruptions via Self-Consistent Stochastic Interpolants2512.10857v2 · Chirag Modi, Jiequn Han, Eric Vanden-Eijnden et al.2025 · 0 citationsarXiv
Geometric Factual Recall in Transformers2605.12426v1 · Shauli Ravfogel, Gilad Yehudai, Joan Bruna et al.2026 · 0 citationsarXiv
Towards Infinitely Long Neural Simulations: Self-Refining Neural Surrogate Models for Dynamical Systems2603.17750v1 · Qi Liu, Laure Zanna, Joan Bruna2026 · 0 citationsarXiv
Compositional Reasoning with Transformers, RNNs, and Chain of Thought2503.01544v2 · Gilad Yehudai, Noah Amsel, Joan Bruna2025 · 0 citationsarXiv
Data-driven multiscale modeling for correcting dynamical systems2303.17496v2 · Karl Otness, Laure Zanna, Joan Bruna2023 · 3 citationsMachine Learning: Science and Technology. 6 (2025) 045072
On Non-Linear operators for Geometric Deep Learning2207.03485v2 · Gr'egoire Sergeant-Perthuis (OURAGAN), Jakob Maier (DYOGENE), Joan Bruna (CIMS) et al.2022 · 1 citationAdvances in Neural Information Processing Systems 35 (NeurIPS 2022)
Axial Neural Networks for Dimension-Free Foundation Models2510.13665v2 · Hyunsu Kim, Jonggeon Park, Joan Bruna et al.2025 · 1 citationNeurIPS 2025
Emergence of Linear Truth Encodings in Language Models2510.15804v1 · Shauli Ravfogel, Gilad Yehudai, Tal Linzen et al.2025 · 0 citationsarXiv
Propagation of Chaos in One-hidden-layer Neural Networks beyond Logarithmic Time2504.13110v3 · Margalit Glasgow, Denny Wu, Joan Bruna2025 · 0 citationsarXiv
Survey on Algorithms for multi-index models2504.05426v2 · Joan Bruna, Daniel Hsu2025 · 3 citationsarXiv
The Generative Leap: Sharp Sample Complexity for Efficiently Learning Gaussian Multi-Index Models2506.05500v1 · Alex Damian, Jason D. Lee, Joan Bruna2025 · 0 citationsarXiv
Distributional Associations vs In-Context Reasoning: A Study of Feed-forward and Attention Layers2406.03068v2 · Lei Chen, Joan Bruna, Alberto Bietti2024 · 0 citationsarXiv
Stochastic Optimal Control Matching2312.02027v5 · Carles Domingo-Enrich, Jiequn Han, Brandon Amos et al.2023 · 0 citationsarXiv
A new approach to observational cosmology using the scattering transform2006.08561v2 · Sihao Cheng, Yuan-Sen Ting, Brice M'enard et al.2020 · 143 citationsMonthly Notices of the Royal Astronomical Society, Volume 499, Issue 4, 2020, pp. 5902-5914
Posterior Sampling with Denoising Oracles via Tilted Transport2407.00745v1 · Joan Bruna, Jiequn Han2024 · 3 citationsarXiv
Exploiting Linear Structure Within Convolutional Networks for Efficient Evaluation1404.0736v2 · Remi Denton, Wojciech Zaremba, Joan Bruna et al.2014 · 1,469 citationsarXiv
Computational-Statistical Gaps in Gaussian Single-Index Models2403.05529v2 · Alex Damian, Loucas Pillaud-Vivien, Jason D. Lee et al.2024 · 0 citationsarXiv
Neural Galerkin Schemes with Active Learning for High-Dimensional Evolution Equations2203.01360v4 · Joan Bruna, Benjamin Peherstorfer, Eric Vanden-Eijnden2022 · 50 citationsJournal of Computational Physics, Volume 496, 2024
Symmetry Breaking in Symmetric Tensor Decomposition2103.06234v2 · Yossi Arjevani, Joan Bruna, Michael Field et al.2021 · 4 citationsarXiv
On Learning Gaussian Multi-index Models with Gradient Flow2310.19793v2 · Alberto Bietti, Joan Bruna, Loucas Pillaud-Vivien2023 · 0 citationsarXiv
A Neural Collapse Perspective on Feature Evolution in Graph Neural Networks2307.01951v2 · Vignesh Kothapalli, Tom Tirer, Joan Bruna2023 · 4 citationsarXiv
Inverse Dynamics Pretraining Learns Good Representations for Multitask Imitation2305.16985v2 · David Brandfonbrener, Ofir Nachum, Joan Bruna2023 · 3 citationsarXiv
On Single Index Models beyond Gaussian Data2307.15804v2 · Joan Bruna, Loucas Pillaud-Vivien, Aaron Zweig2023 · 1 citationarXiv
Symmetric Single Index Learning2310.02117v1 · Aaron Zweig, Joan Bruna2023 · 0 citationsarXiv
Beyond the Edge of Stability via Two-step Gradient Updates2206.04172v3 · Lei Chen, Joan Bruna2022 · 0 citationsarXiv
Reliable coarse-grained turbulent simulations through combined offline learning and neural emulation2307.13144v1 · Christian Pedersen, Laure Zanna, Joan Bruna et al.2023 · 0 citationsarXiv
Towards Antisymmetric Neural Ansatz Separation2208.03264v3 · Aaron Zweig, Joan Bruna2022 · 1 citationarXiv
Conditionally Strongly Log-Concave Generative Models2306.00181v1 · Florentin Guth, Etienne Lempereur, Joan Bruna et al.2023 · 0 citationsarXiv
On the equivalence between graph isomorphism testing and function approximation with GNNs1905.12560v2 · Zhengdao Chen, Soledad Villar, Lei Chen et al.2019 · 126 citationsOriginal version published at the 33rd Conference on Neural Information Processing Systems (NeurIPS 2019)
When does return-conditioned supervised learning work for offline reinforcement learning?2206.01079v3 · David Brandfonbrener, Alberto Bietti, Jacob Buckman et al.2022 · 7 citationsarXiv
Exponential Separations in Symmetric Neural Networks2206.01266v3 · Aaron Zweig, Joan Bruna2022 · 1 citationarXiv
Learning Single-Index Models with Shallow Neural Networks2210.15651v1 · Alberto Bietti, Joan Bruna, Clayton Sanford et al.2022 · 13 citationsarXiv
Cartoon Explanations of Image Classifiers2110.03485v5 · Stefan Kolek, Duc Anh Nguyen, Ron Levie et al.2021 · 13 citationsarXiv
Extended Unconstrained Features Model for Exploring Deep Neural Collapse2202.08087v3 · Tom Tirer, Joan Bruna2022 · 7 citationsarXiv
A Functional Perspective on Learning Symmetric Functions with Neural Networks2008.06952v4 · Aaron Zweig, Joan Bruna2020 · 1 citationarXiv
Kymatio: Scattering Transforms in Python1812.11214v3 · Mathieu Andreux, Tom'as Angles, Georgios Exarchakis et al.2018 · 94 citationsarXiv
On Feature Learning in Neural Networks with Global Convergence Guarantees2204.10782v1 · Zhengdao Chen, Eric Vanden-Eijnden, Joan Bruna2022 · 1 citationarXiv
Backplay: "Man muss immer umkehren"1807.06919v5 · Cinjon Resnick, Roberta Raileanu, Sanyam Kapoor et al.2018 · 23 citationsarXiv
Pommerman: A Multi-Agent Playground1809.07124v2 · Cinjon Resnick, Wes Eldridge, David Ha et al.2018 · 2 citationsarXiv
A Dynamical Central Limit Theorem for Shallow Neural Networks2008.09623v3 · Zhengdao Chen, Grant M. Rotskoff, Joan Bruna et al.2020 · 1 citationarXiv
Simultaneous Transport Evolution for Minimax Equilibria on Measures2202.06460v2 · Carles Domingo-Enrich, Joan Bruna2022 · 0 citationsarXiv
Dual Training of Energy-Based Models with Overparametrized Shallow Neural Networks2107.05134v2 · Carles Domingo-Enrich, Alberto Bietti, Marylou Gabri'e et al.2021 · 1 citationarXiv
Lattice-Based Methods Surpass Sum-of-Squares in Clustering2112.03898v2 · Ilias Zadik, Min Jae Song, Alexander S. Wein et al.2021 · 2 citationsarXiv
Offline RL Without Off-Policy Evaluation2106.08909v3 · David Brandfonbrener, William F. Whitney, Rajesh Ranganath et al.2021 · 16 citationsarXiv
Quantile Filtered Imitation Learning2112.00950v1 · David Brandfonbrener, William F. Whitney, Rajesh Ranganath et al.2021 · 0 citationsarXiv
Career total: 187 works. 119 are in this corpus.Showing the 50 most recent.
Profile built from the corpus for this byline.
Author records are still filling in while the Hub is in alpha. If this is your page, you'll be able to claim it soon. Spot a mistake? Tell us.