PeopleofSuperintelligence

Researcher

Atsushi Nitanda

Atsushi Nitanda is a machine-learning researcher whose work connects optimization theory with neural networks and generative models. He earned degrees in mathematics and information science at Sophia University and the University of Tokyo, and his career includes roles at NTT DATA Mathematical Systems, the University of Tokyo and Kyushu Institute of Technology. His research ranges from accelerated stochastic methods and Boltzmann-machine training to neural-network theory; a 2021 paper with Taiji Suzuki won an ICLR Outstanding Paper Award. In 2025, he coauthored a distribution-optimization method for diffusion-model alignment, and as of October 2026 he is Principal Scientist and Group Lead at A*STAR CFAR and an Associate Professor at NTU.

Timeline

  1. 2007–2018

    Mathematics and a PhD

    He earns a BSc in Mathematics at Sophia University in 2007 and an MSc in Mathematical Sciences at the University of Tokyo in 2009. He completes a PhD in Information Science and Technology at the University of Tokyo in 2018.

  2. 2014

    Accelerated stochastic optimization

    In 2014, he proposes an accelerated mini-batch method that combines Nesterov acceleration with variance reduction for stochastic gradients.

  3. 2017

    Training Boltzmann machines

    With Taiji Suzuki, he proposes a stochastic difference-of-convex algorithm for learning Boltzmann machines in 2017. Their experiments train binary restricted and deep Boltzmann machines without pre-training.

  4. 2021

    Averaged SGD and random features

    With Taiji Suzuki, he shows that averaged stochastic gradient descent can reach the minimax optimal convergence rate for overparameterized two-layer neural networks under the neural tangent kernel regime. Their ICLR 2021 paper wins the Outstanding Paper Award. With Shingo Yashima and Taiji Suzuki, he studies stochastic gradient descent with random features in 2021; their work shows exponential convergence of classification error under a strong low-noise condition.

  5. 2021

    Particle dual averaging

    At NeurIPS 2021, he and Denny Wu and Taiji Suzuki propose particle dual averaging to optimize two-layer mean-field neural networks over probability distributions, with quantitative runtime guarantees.

  6. By 2023

    Faculty posts in Japan

    By 2021, he is affiliated with Kyushu Institute of Technology and RIKEN AIP. In 2023, he is an Associate Professor at Kyushu; his career also includes an Assistant Professor post at the University of Tokyo and research at NTT DATA Mathematical Systems.

  7. 2025

    Aligning diffusion models

    With Ryotaro Kawata, Kazusato Oko and Taiji Suzuki, he introduces a diffusion-model alignment method in 2025. It frames the task as regularized loss minimization over probability distributions and optimizes the distribution with dual averaging.

  8. As of October 2026

    A*STAR and NTU roles

    As of October 2026, he is Principal Scientist and Group Lead at A*STAR CFAR, where he leads the Optimization in AI group, and Associate Professor at NTU's College of Computing and Data Science.

Career

  1. NowAgency for Science, Technology and ResearchPrincipal Scientist and Group Lead
  2. NowNanyang Technological UniversityAssociate Professor
  3. University of TokyoAssistant Professor
  4. Kyushu Institute of TechnologyAssociate Professor

Education

  1. 2018University of TokyoPhD in Information Science and Technology
  2. 2009University of TokyoMSc in Mathematical Sciences
  3. 2007Sophia UniversityBSc in Mathematics

From the biography.

  1. The Massachusetts Institute of Technology trained or employed 44 of the 187 AI major-prize winners, more than any other universityIncludes Atsushi Nitanda. The Massachusetts Institute of Technology trained or employed 44 of the 187 AI major-prize winners, more than any other university. We compare doctorate and 5+ year employment links, career timing and current workplaces for 187 major-prize and 1,111 best-paper winners.
  2. More people shaping AI studied at Massachusetts Institute of Technology than at any other institutionIncludes Nanyang Technological University. We estimate where the people shaping AI studied and where their careers led. Massachusetts Institute of Technology leads with 467 former students, including 291 recorded doctorates.
  3. The company share of AI researchers’ top papers was 31% in 2020–2024 and 8% in 1970–1974Includes Nanyang Technological University. We estimate where the people shaping AI worked when their most-cited papers appeared: the company share was 31% in 2020–2024 and 8% in 1970–1974, and Google led all organizations in 2020–2024.
  4. Google has the highest combined h-index among AI organizationsIncludes Nanyang Technological University. We estimate that Google holds the highest combined h-index at 8,016, while universities account for 73% of the deduplicated total across AI organizations.

Colleagues

People named in this biography, and people with dated positions at the same organization in the same years.

Sources and credits

Photo: We have no published photo of Atsushi Nitanda yet. Send us a free-licensed photo.

  1. Meet Our New Faculty | CCDS NTU Singapore | College of Computing and Data Science | NTU Singapore ntu.edu.sg. Checked 21 October 2026.
  2. Stochastic Proximal Gradient Descent with Acceleration Techniques papers.nips.cc. Checked 21 October 2026.
  3. Stochastic Difference of Convex Algorithm and its Application to Training Deep Boltzmann Machines proceedings.mlr.press. Checked 21 October 2026.
  4. [2006.12297] Optimal Rates for Averaged Stochastic Gradient Descent under Neural Tangent Kernel Regime arxiv.org. Checked 21 October 2026.
  5. Deep Learning Theory Team | RIKEN riken.jp. Checked 21 October 2026.
  6. Exponential Convergence Rates of Classification Errors on Learning with SGD and Random Features proceedings.mlr.press. Checked 21 October 2026.
  7. Particle Dual Averaging: Optimization of Mean Field Neural Network with Global Convergence Rate Analysis papers.nips.cc. Checked 21 October 2026.
  8. 17 papers were accepted at NeurIPS 2021 | Center for Advanced Intelligence Project (AIP) aip.riken.jp. Checked 21 October 2026.
  9. Optimisation Theory for Neural Networks and Implicit Bias Towards Flat Minima a-star.edu.sg. Checked 21 October 2026.
  10. Direct Distributional Optimization for Provable Alignment of Diffusion Models proceedings.iclr.cc. Checked 21 October 2026.
  11. Faculty Directory | College of Computing and Data Science | NTU Singapore ntu.edu.sg. Checked 21 October 2026.
  12. Atsushi Nitanda sites.google.com. Checked 21 October 2026.