Sotabase

Career

· DeepMind
· ETH Postdoctoral Fellow, ETH Zurich
· Software Engineer, Google DeepMind
· R&D Engineer I, Subaru of America
· PhD in Computer Science, Tokyo Institute of Technology
· M.S.E. in Mechanical Engineering, University of Michigan

Publications (15)

Practical Deep Learning with Bayesian Principles
Neural Information Processing Systems · 2019
267
cited
Large-Scale Distributed Second-Order Optimization Using Kronecker-Factored Approximate Curvature for Deep Convolutional Neural Networks
Computer Vision and Pattern Recognition · 2018
96
cited
Efficient Quantized Sparse Matrix Operations on Tensor Cores
International Conference for High Performance Computing, Networking, Storage and Analysis · 2022
49
cited
Scalable and Practical Natural Gradient for Large-Scale Deep Learning
IEEE Transactions on Pattern Analysis and Machine Intelligence · 2020
44
cited
Second-order Optimization Method for Large Mini-batch: Training ResNet-50 on ImageNet in 35 Epochs
arXiv.org · 2018
38
cited
PipeFisher: Efficient Training of Large Language Models Using Pipelining and Fisher Information Matrices
Conference on Machine Learning and Systems · 2022
36
cited
Understanding approximate Fisher information for fast convergence of natural gradient descent in wide neural networks
Neural Information Processing Systems · 2020
31
cited
ASDL: A Unified Interface for Gradient Preconditioning in PyTorch
arXiv.org · 2023
19
cited
Understanding Gradient Regularization in Deep Learning: Efficient Finite-Difference Computation and Implicit Bias
International Conference on Machine Learning · 2022
19
cited
Accelerating Matrix Multiplication in Deep Learning by Using Low-Rank Approximation
International Symposium on High Performance Computing Systems and Applications · 2017
18
cited
Rich Information is Affordable: A Systematic Performance Analysis of Second-order Optimization Using K-FAC
Knowledge Discovery and Data Mining · 2020
18
cited
Neural Graph Databases
LOG IN · 2022
11
cited
Improving Continual Learning by Accurate Gradient Reconstructions of the Past
Trans. Mach. Learn. Res. · 2023
5
cited
Performance Optimizations and Analysis of Distributed Deep Learning with Approximated Second-Order Optimization Method
ICPP Workshops · 2019
5
cited
Evaluating the Compression Efficiency of the Filters in Convolutional Neural Networks
International Conference on Artificial Neural Networks · 2017
4
cited
Sotabase
Kazuki Osawa | Researcher Profile | Sotabase | Sotabase