Research Experience

• 2024.04 - present: Chief Research Scientist at SB Intuitions Corp.

  • 2025.01 - present: Director of Multimodal AI department

  • 2024.11 - 2024.12: Manager of Vision & Language team

  • 2024.04 - 2024.10: Senior Research Scientist at SB Intuitions Corp.

  • 2024.04 - 2024.09: Manager at SoftBank Corp.

• 2018.04 - 2024.03: Research Scientist at Denso IT Laboratory, Inc.

• 2016.05 - 2018.03: Research assistant at National Institute of Advanced Industrial Science and Technology (AIST)

Education

• 2019.04 - 2022.03: Ph.D. in Engineering at Graduate School of Science and Technology, Keio University

• 2016.04 - 2018.03: Master of Science in Engineering at Graduate School of Science and Technology, Keio University

• 2012.04 - 2016.03: Bachelor of Engineering at Department of Electronics and Electrical Engineering, Faculty of Science and Technology, Keio University

Publications

• Graph-Text Contrastive Learning of Inorganic Crystal Structure toward a Foundation Model of Inorganic Materials
  Keisuke Ozawa, Teppei Suzuki, Shunsuke Tonogai, Tomoya Itakura
  Science and Technology of Advanced Materials, 2024, paper, ChemRxiv 

• Fed3DGS: Scalable 3D Gaussian Splatting with Federated Learning
  Teppei Suzuki
  arXiv:2403.11460, arXiv, code

• Federated Learning for Large-Scale Scene Modeling with Neural Radiance Fields
  Teppei Suzuki
  arXiv:2309.06030, arXiv

• Multi-task Curriculum Learning based on Gradient Similarity
  Hiroaki Igarashi, Kenichi Yoneji, Kohta Ishikawa, Rei Kawakami, Teppei Suzuki, Shingo Yashima, Ikuro Sato
  BMVC2022, pdf 

• Feature Space Particle Inference for Neural Network Ensembles
  Shingo Yahima, Teppei Suzuki, Kohta Ishikawa, Ikuro Sato, Rei Kawakami
  ICML2022, PMLR, arXiv, code

• TeachAugment: Data Augmentation Optimization Using Teacher Knowledge
  Teppei Suzuki
  CVPR2022 (Oral), CVF, arXiv, code 

• Clustering as Attention: Unified Image Segmentation with Hierarchical Clustering
  Teppei Suzuki
  arXiv:2205.09949, arXiv, code

• Rethinking PointNet Embedding for Faster and Compact Model
  Teppei Suzuki, Keisuke Ozawa, Yusuke Sekikawa
  3DV2020, arXiv, IEEEXplore 

• Superpixel Segmentation via Convolutional Neural Networks with Regularized Information Maximization
  Teppei Suzuki
  ICASSP2020, IEEEXplore, arXiv, code

• Unsupervised Auto-Encoding Multiple-Object Tracker for constraint-consistent combinatorial problem
  Yuta Kawachi and Teppei Suzuki
  ICASSP2020, IEEEXplore

• Joint Pedestrian Detection and Risk-level Prediction with Motion-Representation-by-Detection
  Hirokatsu Kataoka, Teppei Suzuki, Kodai Nakashima, Yutaka Satoh, Yoshimitsu Aoki
  ICRA2020, IEEEXplore 

• QR-code Reconstruction from Event Data via Optimization in Code Subspace
  Jun Nagata, Yusuke Sekikawa, Teppei Suzuki, Yoshimitsu Aoki
  WACV2020, CVF

• Adversarial Transformations for Semi-Supervised Learning
  Teppei Suzuki and Ikuro Sato
  AAAI2020 arXiv, pdf 

• Superpixel Convolution for Segmentation
  Teppei Suzuki,  Shuichi Akizuki, Naoki Kato, Yoshimitsu Aoki
  ICIP2018, IEEEXplore 

• Drive Video Analysis for the Detection of Traffic Near-Miss Incidents
  Hirokatsu Kataoka, Teppei Suzuki, Shoko Oikawa, Yasuhiro Matsui, Yutaka Satoh
  ICRA2018, arXiv, IEEEXplore 

Award

• MIRU2021, インタラクティブ発表賞, link 

• SSII2021, インタラクティブセッション・オーディエンス賞, link 

• MIRU2017, 学生優秀賞, link 

Talks

• 大規模言語モデルの開発, JSAI2025, slide(ja) 

• 大規模視覚言語モデルによる画像の認識と理解, PRMU2025.03.18

• TeachAugment: Data Augmentation Optimization Using Teacher Knowledge, MIRU2022

• 企業研究所における個人重視の組織体制, PRMU2021.10.08

• Adversarial Transformations for Semi-Supervised Learning, MIRU2020

• 深層学習における半教師あり学習の最新動向, SSII2020, slide(ja)

Others

• Developing large vision-language models

  • Sarashina2-Vision 8B, Sarashina2-Vision 14B, Tech Blog (Model), Tech Blog (Evaluation)

• Reviewer, CVPR/ECCV/ICCV/WACV/NeurIPS/ICLR/ICML etc.

• チュートリアル委員, SSII2024, link 

• PRMU/CVIM研究メンターシッププログラム 2023/2024, link

• シニア評価委員 / Senior  Evaluators, MIRU2022 - 2024, link