About

I’m Shingo Yashima (八嶋晋吾), a machine learning researcher at Denso IT Laboratory.
My research interests include machine learning, statistical learning theory, and computer vision.

Biography

Apr. 2020 - present: Researcher, Denso IT Laboratory.: Research and development on machine learning, especially for AD/ADAS systems.

Mar. 2020: Master of Information Science and Technology, The University of Tokyo.: Supervisor: Taiji Suzuki; Thesis: “Generalization Error Analysis of Stochastic Gradient Descent on Classification Problems under Low Noise Condition” [pdf]

Mar. 2018: Bachelor of Engineering, The University of Tokyo.: Supervisor: Kensuke Aishima; Thesis (in Japanese): "動的モード分解を用いたモデル縮減"

Publications

International Conferences/Workshops (refereed)

Tomoya Takahashi, Shingo Yashima, Kohta Ishikawa, Ikuro Sato, Rio Yokota. Pixel-level Contrastive Learning of Driving Videos with Optical Flow. The 1st Workshop on End-to-End Autonomous Driving (CVPR Workshop), 2023.
Yusuke Sekikawa, Shingo Yashima. Bit-Pruning: A Sparse Multiplication-Less Dot-Product. The 11th International Conference on Learning Representations (ICLR), 2023. [paper] [code]
Hiroaki Igarashi, Kenichi Yoneji, Kohta Ishikawa, Rei Kawakami, Teppei Suzuki, Shingo Yashima, Ikuro Sato. Multi-task Curriculum Learning based on Gradient Similarity. The 33rd British Machine Vision Conference (BMVC), 2022. [paper]
Shingo Yashima. Effectiveness of Function Matching in Driving Scene Recognition. The 3rd Autonomous Vehicle Vision Workshop (ECCV Workshop), 2022. [paper]
Shingo Yashima, Teppei Suzuki, Kohta Ishikawa, Ikuro Sato, Rei Kawakami. Feature Space Particle Inference for Neural Network Ensembles. The 39th International Conference on Machine Learning (ICML), 2022. [paper] [code]
Shingo Yashima, Atsushi Nitanda, Taiji Suzuki. Exponential Convergence Rates of Classification Errors on Learning with SGD and Random Features. The 24th International Conference on Artificial Intelligence and Statistics (AISTATS), 2021. [paper]

Domestic Conferences (non-refereed)

Shingo Yashima, Teppei Suzuki, Kohta Ishikawa, Ikuro Sato, and Rei Kawakami. Feature Space Particle Inference for Neural Network Ensembles. 画像の認識・理解シンポジウム (MIRU), 2022.
高橋那弥, 八嶋晋吾, 石川康太, 佐藤育郎, 横田理央. 走行動画の大規模自己教師あり学習の検討と計画. 画像の認識・理解シンポジウム (MIRU), 2022.
五十嵐博昭, 米司健一, 石川康太, 川上玲, 鈴木哲平, 八嶋晋吾, 佐藤育郎. 勾配の類似度を用いたマルチタスクカリキュラム学習. 画像の認識・理解シンポジウム (MIRU), 2022.
八嶋晋吾. 画像外観検査における異常の種類にロバストな検知手法．情報論的学習理論ワークショップ (IBIS), 2021.
八嶋晋吾, 二反田篤史, 鈴木大慈．Random Featureを用いた確率的勾配法の期待識別誤差の収束解析．情報論的学習理論ワークショップ (IBIS), 2019. 学生優秀プレゼンテーション賞ファイナリスト.
八嶋晋吾, 二反田篤史, 鈴木大慈．カーネル法におけるRandom Featureを用いた確率的勾配法の期待識別誤差の線形収束性．統計関連学会連合大会, 2019.

Talks

セーフティクリティカルな応用のための機械学習モデルの品質管理の考え方. 東京工業大学社会的課題解決型データサイエンス・AI研究推進体シンポジウム, 2022. [program]

Media

デンソーITラボが新アンサンブル学習技術、ベイジアンニューラルネットで最高水準の精度や頑健性. 日経Robotics, 2023. [link]