v1v2v3 (latest)

Hardness of Noise-Free Learning for Two-Hidden-Layer Neural Networks

Neural Information Processing Systems (NeurIPS), 2022

10 February 2022

Papers citing "Hardness of Noise-Free Learning for Two-Hidden-Layer Neural Networks"

22 / 22 papers shown

Title
Learning Neural Networks with Distribution Shift: Efficiently Certifiable GuaranteesInternational Conference on Learning Representations (ICLR), 2025 Gautam Chandrasekaran Adam R. Klivans Lin Lin Lee Konstantinos Stavropoulos OOD 166 1 0 22 Feb 2025
On the Hardness of Learning One Hidden Layer Neural Networks Shuchen Li Ilias Zadik Manolis Zampetakis 134 2 0 04 Oct 2024
Sequencing the Neurome: Towards Scalable Exact Parameter Reconstruction of Black-Box Neural Networks Judah Goldfeder Quinten Roets Gabe Guo John Wright Hod Lipson 165 1 0 27 Sep 2024
Computability of Classification and Deep Learning: From Theoretical Limits to Practical Feasibility through QuantizationJournal of Fourier Analysis and Applications (JFAA), 2024 Holger Boche Vít Fojtík Adalbert Fono Gitta Kutyniok 302 1 0 12 Aug 2024
Learning Neural Networks with Sparse Activations Pranjal Awasthi Nishanth Dikkala Pritish Kamath Raghu Meka 327 5 0 26 Jun 2024
Hardness of Learning Neural Networks under the Manifold Hypothesis B. Kiani Jason Wang Melanie Weber 199 11 0 03 Jun 2024
Exploration is Harder than Prediction: Cryptographically Separating Reinforcement Learning from Supervised LearningIEEE Annual Symposium on Foundations of Computer Science (FOCS), 2024 Noah Golowich Ankur Moitra Dhruv Rohatgi OffRL 168 6 0 04 Apr 2024
Provably learning a multi-head attention layer Sitan Chen Yuanzhi Li MLT 214 17 0 06 Feb 2024
The Benefits of Reusing Batches for Gradient Descent in Two-Layer Networks: Breaking the Curse of Information and Leap ExponentsInternational Conference on Machine Learning (ICML), 2024 Yatin Dandi Emanuele Troiani Luca Arnaboldi Luca Pesce Lenka Zdeborová Florent Krzakala MLT 246 36 0 05 Feb 2024
Looped Transformers are Better at Learning Learning AlgorithmsInternational Conference on Learning Representations (ICLR), 2023 Liu Yang Kangwook Lee Robert D. Nowak Dimitris Papailiopoulos 354 53 0 21 Nov 2023
Polynomial-Time Solutions for ReLU Network Training: A Complexity Classification via Max-Cut and Zonotopes Yifei Wang Mert Pilanci 200 4 0 18 Nov 2023
Optimizing Solution-Samplers for Combinatorial Problems: The Landscape of Policy-Gradient MethodsNeural Information Processing Systems (NeurIPS), 2023 Constantine Caramanis Eleni Psaroudaki Alkis Kalavasis Vasilis Kontonis Christos Tzamos 250 5 0 08 Oct 2023
Efficiently Learning One-Hidden-Layer ReLU Networks via Schur PolynomialsAnnual Conference Computational Learning Theory (COLT), 2023 Ilias Diakonikolas D. Kane 143 6 0 24 Jul 2023
Most Neural Networks Are Almost LearnableNeural Information Processing Systems (NeurIPS), 2023 Amit Daniely Nathan Srebro Gal Vardi 176 0 0 25 May 2023
Algorithmic Decorrelation and Planted Clique in Dependent Random Graphs: The Case of Extra TrianglesIEEE Annual Symposium on Foundations of Computer Science (FOCS), 2023 Guy Bresler Chenghao Guo Yury Polyanskiy 162 3 0 17 May 2023
Computational Complexity of Learning Neural Networks: Smoothness and DegeneracyNeural Information Processing Systems (NeurIPS), 2023 Amit Daniely Nathan Srebro Gal Vardi 198 6 0 15 Feb 2023
Learning Single-Index Models with Shallow Neural NetworksNeural Information Processing Systems (NeurIPS), 2022 A. Bietti Joan Bruna Clayton Sanford M. Song 288 83 0 27 Oct 2022
Magnitude and Angle Dynamics in Training Single ReLU NeuronsNeural Networks (NN), 2022 Sangmin Lee Byeongsu Sim Jong Chul Ye MLT 303 6 0 27 Sep 2022
Sampling is as easy as learning the score: theory for diffusion models with minimal data assumptionsInternational Conference on Learning Representations (ICLR), 2022 Sitan Chen Sinho Chewi Jungshian Li Yuanzhi Li Adil Salim Anru R. Zhang DiffM 468 347 0 22 Sep 2022
Learning (Very) Simple Generative Models Is HardNeural Information Processing Systems (NeurIPS), 2022 Sitan Chen Jungshian Li Yuanzhi Li 140 11 0 31 May 2022
Learning ReLU networks to high uniform accuracy is intractableInternational Conference on Learning Representations (ICLR), 2022 Julius Berner Philipp Grohs F. Voigtlaender 184 5 0 26 May 2022
$Training Fully Connected Neural Networks is $\exists\mathbb{R}$-Complete$ Training Fully Connected Neural Networks is $\exists\mathbb{R}$ -CompleteNeural Information Processing Systems (NeurIPS), 2022 Daniel Bertschinger Christoph Hertrich Paul Jungeblut Tillmann Miltzow Simon Weber OffRL 250 35 0 04 Apr 2022