He received his B.E. and M.E. degrees in Physics and Physico-Informatics Engineering from Keio University, Japan, in 2013, 2015, respectively. He recieved the Ph.D. degree in Engineering from Keio University, Japan in 2020. He is a researcher of NTT since 2015.
Publications
Journal
- Sekitoshi Kanai, Masanori Yamada, Hiroshi Takahashi, Yuki Yamanaka, Yasutoshi Ida,
Relationship between Non-smoothness in Adversarial Training, Constraints of Attacks, and Flatness in the Input Space,
IEEE TNNLS Vol. 35, No. 8, pp. 10817-10831, (2024)
[paper], [preprint (arXiv)] - Yasuhiro Fujiwara, Sekitoshi Kanai, Yasutoshi Ida, Atsutoshi Kumagai, Naonori Ueda,
Fast Algorithm for Anchor Graph Hashing, PVLDB2021, Vol. 14, No. 6, pp.916-928, (2021)
[paper] - Sekitoshi Kanai, Maho Sugaya, Shuichi Adachi, Kentaro Matsui,
Low-complexity simultaneous estimation of head-related transfer functions by prediction error method,
Journal of the Audio Engineering Society, Vol.64, No.11, pp.895-904, (2016)
[paper]
International Conference
- Yasutoshi Ida, Sekitoshi Kanai, Atsutoshi Kumagai, Tomoharu Iwata, Yasuhiro Fujiwara,
Fast Iterative Hard Thresholding Methods with Pruning Gradient Computations, NeurIPS 2024 (To appear)
[paper], - Shin’ya Yamaguchi, Sekitoshi Kanai, Kazuki Adachi, Daiki Chijiwa,
Adaptive Random Feature Regularization on Fine-tuning Deep Neural Networks, CVPR 2024
[paper], [arXiv] - Shin’ya Yamaguchi, Daiki Chijiwa, Sekitoshi Kanai, Atsutoshi Kumagai, Hisashi Kashima
Regularizing Neural Networks with Meta-Learning Generative Models, NeurIPS 2023
[paper], [arXiv] - Satoshi Suzuki, Shin’ya Yamaguchi, Shoichiro Takeda, Sekitoshi Kanai, Naoki Makishima, Atsushi Ando, Ryo Masumura
Adversarial Finetuning with Latent Representation Constraint to Mitigate Accuracy-Robustness Tradeoff, ICCV 2023
[paper], [arXiv] - Sekitoshi Kanai, Shin’ya Yamaguchi, Masanori Yamada, Hiroshi Takahashi, Kentaro Ohno, Yasutoshi Ida,
One-vs-the-Rest Loss to Focus on Important Samples in Adversarial Training, ICML 2023
[paper], [video & poster], [arXiv] - Yasutoshi Ida, Sekitoshi Kanai, Atsutoshi Kumagai
Fast Block Coordinate Descent for Non-Convex Group Regularizations, AISTATS 2023
[paper] - Kentaro Ohno, Sekitoshi Kanai, Yasutoshi Ida,
Fast Saturating Gate for Learning Long Time Scales with Recurrent Neural Networks, AAAI 2023
[paper], [arXiv] - Yasutoshi Ida, Sekitoshi Kanai, Kazuki Adachi, Atsutoshi Kumagai, Yasuhiro Fujiwara,
Fast Regularized Discrete Optimal Transport with Group-sparse Regularizers, AAAI 2023
[paper], [arXiv] - Hiroshi Takahashi, Tomoharu Iwata, Atsutoshi Kumagai, Sekitoshi Kanai, Masanori Yamada, Yuuki Yamanaka, and Hisashi Kashima,
Learning Optimal Priors for Task-Invariant Representations in Variational Autoencoders, KDD2022
[paper] - Yasuhiro Fujiwara, Yasutoshi Ida, Atsutoshi Kumagai, Sekitoshi Kanai, and Naonori Ueda,
Fast and Accurate Anchor Graph-based Label Prediction, CIKM2021
[paper] - Shin’ya Yamaguchi, Sekitoshi Kanai,
F-Drop&Match: GANs with a Dead Zone in the High-Frequency Domain, ICCV 2021
[paper], [arXiv] - Shin’ya Yamaguchi, Sekitoshi Kanai, Tetsuya Shioda, Shoichiro Takeda,
Multiple Pretext-Task for Self-Supervised Learning via Mixing Multiple Image Transformations, ICIP2021
[paper], [arxiv] - Sekitoshi Kanai, Masanori Yamada, Shin’ya Yamaguchi, Hiroshi Takahashi, Yasutoshi Ida,
Constraining Logits by Bounded Function for Adversarial Robustness, IJCNN2021
[paper], [arxiv] - Toshiaki Wakatsuki, Sekitoshi Kanai, Yasuhiro Fujiwara,
Accelerate Inference of CNNs for Video Analysis While Preserving Exactness Exploiting Activation Sparsity, MLSys2021
[paper] - Yasuhiro Fujiwara, Yasutoshi Ida, Sekitoshi Kanai, Atsutoshi Kumagai, Naonori Ueda,
Fast Similarity Computation for t-SNE, ICDE2021
[paper] - Yasuhiro Fujiwara, Atsutoshi Kumagai, Sekitoshi Kanai, Yasutoshi Ida, and Naonori Ueda,
Efficient Algorithm for the b-Matching, KDD2020
[paper] - Yasutoshi Ida, Sekitoshi Kanai,Yasuhiro Fujiwara, Tomoharu Iwata, Koh Takeuchi, and Hisashi Kashima,
Fast Deterministic CUR Matrix Decomposition with Accuracy Assurance, ICML2020,
[paper] - Sekitoshi Kanai, Yasutoshi Ida, Yasuhiro Fujiwara, Masanori Yamada, Shuichi Adachi,
Absum: Simple Regularization Method for Reducing Structural Sensitivity of Convolutional Neural Networks, AAAI2020
[paper], [arxiv] - Shin’ya Yamaguchi, Sekitoshi Kanai, Takeharu Eda,
Effective Data Augmentation with Multi-Domain Learning GANs, AAAI 2020
[paper], [arxiv] - Yasuhiro Fujiwara, Yasutoshi Ida, Sekitoshi Kanai, Atsutoshi Kumagai, Junya Arai, Naonori Ueda,
Fast Random Forest Algorithm via Incremental Upper Bound, CIKM 2019
[paper] - Yuki Yamanaka, Tomoharu Iwata, Hiroshi Takahashi, Masanori Yamada, Sekitoshi Kanai,
Autoencoding Binary Classifiers for Supervised Anomaly Detection, PRICAI2019
[paper] - Yasuhiro Fujiwara, Sekitoshi Kanai, Junya Arai, Yasutoshi Ida, Naonori Ueda,
Efficient Data Point Pruning for One-Class SVM, AAAI2019
[paper] - Yasuhiro Fujiwara, Junya Arai, Sekitoshi Kanai, Yasutoshi Ida, Naonori Ueda,
Adaptive Data Pruning for Support Vector Machines, 2018 IEEE International Conference on Big Data
[paper] - Sekitoshi Kanai, Yasuhiro Fujiwara, Yuki Yamanaka, Shuichi Adachi,
Sigsoftmax: Reanalysis of the Softmax Bottleneck, NeurIPS 2018
[paper], [arxiv] - Sekitoshi Kanai, Yasuhiro Fujiwara, Sotetsu Iwamura,
Preventing Gradient Explosions in Gated Recurrent Units, NIPS 2017
[paper] - Sekitoshi Kanai, Kentaro Matsui, Yasushige Nakayama, Shuichi Adachi,
Uncorrelated Input Signals Design and Identification with Low-Complexity for Simultaneous Estimation of Head-Related Transfer Functions, 137th Audio Engineering Society Convention (2014)
[paper] - Sekitoshi Kanai, Kentaro Matsui, Shuichi Adachi,
Identification input design for simultaneous estimation of head-related transfer functions, SICE2014
[paper]
Preprints
- Sekitoshi Kanai, Yasutoshi Ida, Kazuki Adachi, Mihiro Uchida, Tsukasa Yoshida, Shin’ya Yamaguchi,
Evaluating Time-Series Training Dataset through Lens of Spectrum in Deep State Space Models
arXiv:2408.16261, 2024.
[arXiv] - Shin’ya Yamaguchi, Sekitoshi Kanai, Atsutoshi Kumagai, Daiki Chijiwa, Hisashi Kashima
Transfer Learning with Pre-trained Conditional Generative Models
arXiv:2204.12833, 2022.
[arXiv] - Masanori Yamada, Sekitoshi Kanai, Tomoharu Iwata, Tomokatsu Takahashi, Yuki Yamanaka, Hiroshi Takahashi, Atsutoshi Kumagai,
Adversarial Training Makes Weight Loss Landscape Sharper in Logistic Regression,
arXiv:2102.02950, 2021.
[arXiv]
Activities
Reviewer:
- ICML 2020-2024
- NeurIPS 2019-2024 (Top Reviewers 2022, 2023)
- ICLR 2022, 2024
- AAAI 2021-2023
- IJCAI-PRICAI 2020
- IJCAI 2024
- CVPR 2024
- IEEE TNNLS
- Neurocomputing