Hadi Daneshmand

39 papers A* 12A 3Journal 22Unranked 1

Year	Rank	Type	Title / Venue / Authors
2025	J	`jnl`	Data Generation without Function Estimation. CoRR Hadi Daneshmand, Ashkan Soleymani
2025	J	`jnl`	Linear Transformers Implicitly Discover Unified Numerical Algorithms. CoRR Patrick Lutz, Aditya Gangrade, Hadi Daneshmand, Venkatesh Saligrama
2025	A*	`conf`	Transformers Can Learn Temporal Difference Methods for In-Context Reinforcement Learning. ICLR Jiuqi Wang, Ethan Blaser, Hadi Daneshmand, Shangtong Zhang
2024	J	`jnl`	Provable optimal transport with transformers: The essence of depth and prompt engineering. CoRR Hadi Daneshmand
2024	A*	`conf`	Towards Training Without Depth Limits: Batch Normalization Without Gradient Explosion. ICLR Alexandru Meterez, Amir Joudaki, Francesco Orabona, Alexander Immer, Gunnar Rätsch, Hadi Daneshmand
2024	J	`jnl`	Transformers Learn Temporal Difference Methods for In-Context Reinforcement Learning. CoRR Jiuqi Wang, Ethan Blaser, Hadi Daneshmand, Shangtong Zhang
2023	A*	`conf`	Efficient displacement convex optimization with particle gradient descent. ICML Hadi Daneshmand, Jason D. Lee, Chi Jin
2023	J	`jnl`	Efficient displacement convex optimization with particle gradient descent. CoRR Hadi Daneshmand, Jason D. Lee, Chi Jin
2023	A*	`conf`	On Bridging the Gap between Mean Field and Finite Width Deep Random Multilayer Perceptron with Batch Normalization. ICML Amir Joudaki, Hadi Daneshmand, Francis R. Bach
2023	A*	`conf`	On the impact of activation and normalization in obtaining isometric embeddings at initialization. NeurIPS Amir Joudaki, Hadi Daneshmand, Francis R. Bach
2023	J	`jnl`	On the impact of activation and normalization in obtaining isometric embeddings at initialization. CoRR Amir Joudaki, Hadi Daneshmand, Francis R. Bach
2023	J	`jnl`	Towards Training Without Depth Limits: Batch Normalization Without Gradient Explosion. CoRR Alexandru Meterez, Amir Joudaki, Francesco Orabona, Alexander Immer, Gunnar Rätsch, Hadi Daneshmand
2023	A*	`conf`	Transformers learn to implement preconditioned gradient descent for in-context learning. NeurIPS Kwangjun Ahn, Xiang Cheng, Hadi Daneshmand, Suvrit Sra
2023	J	`jnl`	Transformers learn to implement preconditioned gradient descent for in-context learning. CoRR Kwangjun Ahn, Xiang Cheng, Hadi Daneshmand, Suvrit Sra
2022	J	`jnl`	Entropy Maximization with Depth: A Variational Principle for Random Neural Networks. CoRR Amir Joudaki, Hadi Daneshmand, Francis R. Bach
2022	J	`jnl`	Polynomial-time sparse measure recovery. CoRR Hadi Daneshmand, Francis R. Bach
2021	A*	`conf`	Batch Normalization Orthogonalizes Representations in Deep Random Networks. NeurIPS Hadi Daneshmand, Amir Joudaki, Francis R. Bach
2021	J	`jnl`	Batch Normalization Orthogonalizes Representations in Deep Random Networks. CoRR Hadi Daneshmand, Amir Joudaki, Francis R. Bach
2021	A*	`conf`	Rethinking the Variational Interpretation of Accelerated Optimization Methods. NeurIPS Peiyuan Zhang, Antonio Orvieto, Hadi Daneshmand
2021	A	`conf`	Revisiting the Role of Euler Numerical Integration on Acceleration and Stability in Convex Optimization. AISTATS Peiyuan Zhang, Antonio Orvieto, Hadi Daneshmand, Thomas Hofmann, Roy S. Smith
2021	J	`jnl`	Revisiting the Role of Euler Numerical Integration on Acceleration and Stability in Convex Optimization. CoRR Peiyuan Zhang, Antonio Orvieto, Hadi Daneshmand, Thomas Hofmann, Roy S. Smith
2020	A*	`conf`	Batch normalization provably avoids ranks collapse for randomly initialised deep networks. NeurIPS Hadi Daneshmand, Jonas Moritz Kohler, Francis R. Bach, Thomas Hofmann, Aurélien Lucchi
2020	—	`—`	Optimization for Neural Networks: Quest for Theoretical Understandings. Hadi Daneshmand
2020	J	`jnl`	Theoretical Understanding of Batch-normalization: A Markov Chain Perspective. CoRR Hadi Daneshmand, Jonas Moritz Kohler, Francis R. Bach, Thomas Hofmann, Aurélien Lucchi
2019	A	`conf`	Exponential convergence rates for Batch Normalization: The power of length-direction decoupling in non-convex optimization. AISTATS Jonas Moritz Kohler, Hadi Daneshmand, Aurélien Lucchi, Thomas Hofmann, Ming Zhou, Klaus Neymeyr
2019	A	`conf`	Local Saddle Point Optimization: A Curvature Exploitation Approach. AISTATS Leonard Adolphs, Hadi Daneshmand, Aurélien Lucchi, Thomas Hofmann
2019	J	`jnl`	Mixing of Stochastic Accelerated Gradient Descent. CoRR Peiyuan Zhang, Hadi Daneshmand, Thomas Hofmann
2018	A*	`conf`	Escaping Saddles with Stochastic Gradients. ICML Hadi Daneshmand, Jonas Moritz Kohler, Aurélien Lucchi, Thomas Hofmann
2018	J	`jnl`	Escaping Saddles with Stochastic Gradients. CoRR Hadi Daneshmand, Jonas Moritz Kohler, Aurélien Lucchi, Thomas Hofmann
2018	J	`jnl`	Local Saddle Point Optimization: A Curvature Exploitation Approach. CoRR Leonard Adolphs, Hadi Daneshmand, Aurélien Lucchi, Thomas Hofmann
2018	J	`jnl`	Towards a Theoretical Understanding of Batch Normalization. CoRR Jonas Moritz Kohler, Hadi Daneshmand, Aurélien Lucchi, Ming Zhou, Klaus Neymeyr, Thomas Hofmann
2017	J	`jnl`	Accelerated Dual Learning by Homotopic Initialization. CoRR Hadi Daneshmand, Hamed Hassani, Thomas Hofmann
2016	—	`conf`	Adaptive Newton Method for Empirical Risk Minimization to Statistical Accuracy. NIPS Aryan Mokhtari, Hadi Daneshmand, Aurélien Lucchi, Thomas Hofmann, Alejandro Ribeiro
2016	J	`jnl`	DynaNewton - Accelerating Newton's Method for Machine Learning. CoRR Hadi Daneshmand, Aurélien Lucchi, Thomas Hofmann
2016	J	`jnl`	Estimating Diffusion Networks: Recovery Conditions, Sample Complexity and Soft-thresholding Algorithm. J. Mach. Learn. Res. Manuel Gomez-Rodriguez, Le Song, Hadi Daneshmand, Bernhard Schölkopf
2016	A*	`conf`	Starting Small - Learning with Adaptive Sample Sizes. ICML Hadi Daneshmand, Aurélien Lucchi, Thomas Hofmann
2016	J	`jnl`	Starting Small - Learning with Adaptive Sample Sizes. CoRR Hadi Daneshmand, Aurélien Lucchi, Thomas Hofmann
2014	A*	`conf`	Estimating Diffusion Network Structures: Recovery Conditions, Sample Complexity & Soft-thresholding Algorithm. ICML Hadi Daneshmand, Manuel Gomez-Rodriguez, Le Song, Bernhard Schölkopf
2014	J	`jnl`	Estimating Diffusion Network Structures: Recovery Conditions, Sample Complexity & Soft-thresholding Algorithm. CoRR Hadi Daneshmand, Manuel Gomez-Rodriguez, Le Song, Bernhard Schölkopf