Soham De
Artificial Intelligence · 2015–2023
A researcher in the Science corpus: 15 papers; trusts 2,179 researchers and is trusted by 708 researchers here, as inferred from the citation graph.
Subfields
- Artificial Intelligence — 540 citations into it; #3,286 of 36,961 by trust here, #4,420 by citations
- Computer Vision and Pattern Recognition — 324 citations into it; #2,157 of 21,973 by trust here, #3,692 by citations
- Optimization & Training — 314 citations into it; #522 of 2,605 by trust here, #619 by citations
- Model Architectures — 119 citations into it; #1,365 of 2,288 by trust here, #1,834 by citations
- Computational Mechanics — 66 citations into it
- Representation & Self-Supervised Learning — 40 citations into it
- Scaling & Learning Theory — 33 citations into it; #377 of 429 by trust here, #398 by citations
- Efficiency & Compression — 31 citations into it; #973 of 1,273 by trust here, #782 by citations
Whom Soham De trusts overall
Their strongest outbound trust, strongest first, inferred from their citations.
- Kaiming He
- Alex Krizhevsky
- Geoffrey E. Hinton
- Yann LeCun
- Diederik P. Kingma
- Yoshua Bengio
- Ilya Sutskever
- X. Zhang
- Karen Simonyan
- Christian Szegedy
Who trusts Soham De overall
- Andrew Brock
- Samuel L. Smith
- Stanislav Fort
- Hugo Touvron
- Tom Sander
- Charlie Nash
- Biyi Fang
- Johannes Klicpera
- Quoc V. Le
- Zheng Xu
Papers from Soham De's perspective
Ordered by how strongly the researchers Soham De trusts endorse each paper. Same corpus, a different order for every researcher.
- Deep Residual Learning for Image Recognition (Computer Vision and Pattern Recognition)
- ImageNet classification with deep convolutional neural networks
- Adam: A Method for Stochastic Optimization (International Conference on Learning Representations)
- Attention is All you Need (Neural Information Processing Systems)
- Long Short-Term Memory
- Gradient-based learning applied to document recognition
- ImageNet: A large-scale hierarchical image database
- Very Deep Convolutional Networks for Large-Scale Image Recognition (International Conference on Learning Representations)
- Learning Multiple Layers of Features from Tiny Images
- Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift (International Conference on Machine Learning)
Their papers
- High-Performance Large-Scale Image Recognition Without Normalization (2021) — 638 citations
- Adversarial Robustness through Local Linearization (2019) — 333 citations
- Batch Normalization Biases Residual Blocks Towards the Identity Function in Deep Networks (2020) — 166 citations
- Characterizing signal propagation to close the performance gap in unnormalized ResNets (2021) — 145 citations
- BYOL works even without batch statistics (2020) — 131 citations
- Training Quantized Nets: A Deeper Understanding (2017) — 230 citations
- Unlocking High-Accuracy Differentially Private Image Classification through Scale (2022) — 286 citations
- On the Generalization Benefit of Noise in Stochastic Gradient Descent (2020) — 129 citations
- Batch Normalization Biases Deep Residual Networks Towards Shallow Paths (2020) — 20 citations
- Automated Inference with Adaptive Batches (2017) — 85 citations
- Big Batch SGD: Automated Inference using Adaptive Batch Sizes (2016) — 64 citations
- Convergence Guarantees for RMSProp and ADAM in Non-Convex Optimization and an Empirical Comparison to Nesterov Acceleration (2018) — 120 citations
- Resurrecting Recurrent Neural Networks for Long Sequences (2023) — 516 citations
- Drawing Multiple Augmentation Samples Per Image During Training Efficiently Decreases Test Error (2021) — 36 citations
- Efficient Distributed SGD with Variance Reduction (2015) — 43 citations