HD
Hadi Daneshmand
cs.LGstat.MLcs.AImath.OCcs.CLcs.SImath.MPmath-phmath.STphysics.soc-ph
On Valency
published · living versionsW_h6dd646v·v1 · currentpublished
On Lipschitz Explosion in Deep Neural Networks with Normalization: Consequences for Optimization and Adversarial Robustness
with Ashkan Soleymani, Reyhaneh Hosseinpourkhoshkbari, Patrick Jaillet
1 version
Preprints & journals
24 papers in the corpus · 2014–2026Nash Equilibrium Text: A Game-Theoretic Decoding Framework for Text Generation2610.05817v1 · Alireza Jafari, Arman Adibi, Mohammad Ghavamzadeh et al.2026 · 0 citationsarXiv
Transformers as In-Context Samplers: From Closed-Form Diffusion to Estimation-Free Sampling2609.08981v2 · Arman Adibi, Alireza Jafari, Mohammad Ghavamzadeh et al.2026 · 0 citationsarXiv
Linear Transformers Implicitly Discover Unified Numerical Algorithms2509.19702v1 · Patrick Lutz, Aditya Gangrade, Hadi Daneshmand et al.2025 · 0 citationsarXiv
Data Generation without Function Estimation2507.08239v1 · Hadi Daneshmand, Ashkan Soleymani2025 · 0 citationsarXiv
Provable optimal transport with transformers: The essence of depth and prompt engineering2410.19931v3 · Hadi Daneshmand2024 · 1 citationarXiv
Transformers Can Learn Temporal Difference Methods for In-Context Reinforcement Learning2405.13861v4 · Jiuqi Wang, Ethan Blaser, Hadi Daneshmand et al.2024 · 0 citationsarXiv
Towards Training Without Depth Limits: Batch Normalization Without Gradient Explosion2310.02012v1 · Alexandru Meterez, Amir Joudaki, Francesco Orabona et al.2023 · 0 citationsarXiv
Transformers learn to implement preconditioned gradient descent for in-context learning2306.00297v2 · Kwangjun Ahn, Xiang Cheng, Hadi Daneshmand et al.2023 · 11 citations37th Conference on Neural Information Processing Systems (NeurIPS 2023)
On the impact of activation and normalization in obtaining isometric embeddings at initialization2305.18399v3 · Amir Joudaki, Hadi Daneshmand, Francis Bach2023 · 5 citationsarXiv
Efficient displacement convex optimization with particle gradient descent2302.04753v1 · Hadi Daneshmand, Jason D. Lee, Chi Jin2023 · 0 citationsarXiv
On Bridging the Gap between Mean Field and Finite Width in Deep Random Neural Networks with Batch Normalization2205.13076v3 · Amir Joudaki, Hadi Daneshmand, Francis Bach2022 · 0 citationsarXiv
Polynomial-time Sparse Measure Recovery: From Mean Field Theory to Algorithm Design2204.07879v4 · Hadi Daneshmand, Francis Bach2022 · 0 citationsarXiv
Rethinking the Variational Interpretation of Nesterov's Accelerated Method2107.05040v1 · Peiyuan Zhang, Antonio Orvieto, Hadi Daneshmand2021 · 0 citationsarXiv
Batch Normalization Orthogonalizes Representations in Deep Random Networks2106.03970v1 · Hadi Daneshmand, Amir Joudaki, Francis Bach2021 · 11 citationsarXiv
Revisiting the Role of Euler Numerical Integration on Acceleration and Stability in Convex Optimization2102.11537v1 · Peiyuan Zhang, Antonio Orvieto, Hadi Daneshmand et al.2021 · 0 citationsarXiv
Batch Normalization Provably Avoids Rank Collapse for Randomly Initialised Deep Networks2003.01652v3 · Hadi Daneshmand, Jonas Kohler, Francis Bach et al.2020 · 7 citationsarXiv
Mixing of Stochastic Accelerated Gradient Descent1910.14616v1 · Peiyuan Zhang, Hadi Daneshmand, Thomas Hofmann2019 · 0 citationsarXiv
Exponential convergence rates for Batch Normalization: The power of length-direction decoupling in non-convex optimization1805.10694v3 · Jonas Kohler, Hadi Daneshmand, Aurelien Lucchi et al.2018 · 46 citationsarXiv
Local Saddle Point Optimization: A Curvature Exploitation Approach1805.05751v3 · Leonard Adolphs, Hadi Daneshmand, Aurelien Lucchi et al.2018 · 63 citationsarXiv
Escaping Saddles with Stochastic Gradients1803.05999v2 · Hadi Daneshmand, Jonas Kohler, Aurelien Lucchi et al.2018 · 84 citationsarXiv
Accelerated Dual Learning by Homotopic Initialization1706.03958v1 · Hadi Daneshmand, Hamed Hassani, Thomas Hofmann2017 · 0 citationsarXiv
DynaNewton - Accelerating Newton's Method for Machine Learning1605.06561v1 · Hadi Daneshmand, Aurelien Lucchi, Thomas Hofmann2016 · 3 citationsarXiv
Starting Small -- Learning with Adaptive Sample Sizes1603.02839v2 · Hadi Daneshmand, Aurelien Lucchi, Thomas Hofmann2016 · 34 citationsarXiv
Estimating Diffusion Network Structures: Recovery Conditions, Sample Complexity & Soft-thresholding Algorithm.25932466 · Daneshmand, Hadi, Gomez-Rodriguez, Manuel, Song, Le et al.2021 · 83 citationsJMLR workshop and conference proceedings. 2014;32(2):793-801
Career total: 31 works. 24 are in this corpus.
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.