Published Papers - HACHIYA Hirotaka

Encoder-decoder-based image transformation approach for integrating precipitation forecasts ACML

Hirotaka Hachiya, Yusuke Masumoto, Yuki Mori, Naonori Ueda （Part： Lead author )

Proceedings of the 13th Asian Conference on Machine Learning (ACML2021)     2021.11  [Refereed]

Multi-stream based Marked Point Process

Sujun Hong, Hirotaka Hachiya （Part： Corresponding author )

Proceedings of the 13th Asian Conference on Machine Learning (ACML2021)     2021.11  [Refereed]

Simulation of broad-band ground motions with consistent long-period and short-period components using the Wasserstein interpolation of acceleration envelopes

Tomohisa Okazaki, Hirotaka Hachiya, Asako Iwaki, Takahiro Maeda, Hiroyuki Fujiwara, Naonori Ueda

Geophysical Journal International ( OXFORD UNIV PRESS )  227 ( 1 ) 333 - 349   2021.10  [Refereed]

　View Summary

Practical hybrid approaches for the simulation of broad-band ground motions often combine long-period and short-period waveforms synthesized by independent methods under different assumptions for different period ranges, which at times can lead to incompatible time histories and frequency properties. This study explores an approach that generates consistent broad-hand waveforms using past observation records, under the assumption that long-period waveforms can he obtained from physics-based simulations. Specifically, acceleration envelopes and Fourier amplitude spectra are transformed from long-period to short-period using machine learning methods, and they arc combined to produce a broad-band waveform. To effectively obtain the relationship of high-dimensional envelopes from limited amount of data, we (I) l'onnulate the problem as the conversion of probability distributions, which enables the introduction of a metric known as the Wasserstein distance, and (2) embed pairs of longperiod and short-period envelopes into a common latent space to improve the consistency of the entire waveform. An experimental application to a past earthquake demonstrates that the proposed method exhibits superior performance compared to existing methods as well as neural network approaches. In particular, the proposed method reproduces global properties in the time domain, which confirms the effectiveness of the embedding approach as well as the advantage of the Wasserstein distance as a measure of dissimilarity of the envelopes. This method serves as a novel machine learning approach that maintains consistency both in the time-domain and frequency-domain properties of waveforms.

DOI

Exchangeable Deep Neural Networks for Set-to-Set Matching and Learning

Yuki Saito, Takuma Nakamura, Hirotaka Hachiya, Kenji Fukumizu

Proceedings of the 16th European Conference on Computer Vision (ECCV2020) ( Springer International Publishing )    626 - 646   2020.08  [Refereed]

DOI

Direct Multi-class AUC Maximization for Forecasting Rapidly Intensifying Tropical Cyclones

Hirotaka Hachiya, Shumpei Kurora, Udai Shimada, Naonori Ueda （Part： Lead author )

JpGU-AGU Joint Meeting 2020     2020.07  [Refereed]

Adaptive truncated residuals regression for fine-grained regression problems

Hirotaka Hachiya, Yu Yamamamoto, Kazuhiro Hirahara, Naonori Ueda （Part： Lead author )

Proceedings of the 11th Asian Conference on Machine Learning (ACML2019) ( ACML )    2019.11  [Refereed]

Machine Learning Approach for Adaptive Integration of Multiple Relative Intensity Models toward Improved Earthquake Forecasts in Japan

Hirotaka Hachiya, Kazuro Hirahara, Naonori Ueda

International Union of Geodesy and Geophysics (IUGG2019) ( IUGG )    2019.07  [Refereed]

Broadband ground-motion synthesis using embeddig machine learning

omohisa Okazaki, Hirotaka Hachiya, Naonori Ueda, Asako Iwaki, Takahiro Maeda, Hiroyuki Fujiwara

International Union of Geodesy and Geophysics (IUGG2019)     2019.05  [Refereed]

Machine learning approach for constraining the plausible ranges of frictional parameters on the Philippine Sea plate reproducing the historical sequences of the Nankai megaquakes

Hirotaka Hachiya, Yu Yamamamoto, Kazuhiro Hirahara, Atsushi Takahashi, Naonori Ueda

European Geosciences Union (EGU) ( EGU )    2019.04  [Refereed]

Synthesis of Broadband Ground Motions Using Embedding and Neural Networks

Tomohisa Okazaki, Hirotaka Hachiya, Naonori Ueda, Asako Iwaki, Takahiro Maeda, Hiroyuki Fujiwara

European Geosciences Union (EGU) ( EGU )    2019.04  [Refereed]

3D Faster R-CNNとレーザスキャンとの組み合わせによる特定物体の頑健な距離推定

八谷大岳、射手矢和真、中村恭之

計測自動制御学会論文集   55(1)   2019.01  [Refereed]

Distance estimation with 2.5D anchors and its application to robot navigation

Hirotaka Hachiya, Yuki Saito, Kazuma Iteya, Masaya Nomura, Takayuki Nakamura

ROBOMECH Journal ( Springer Nature )  5 ( 1 )   2018.12  [Refereed]

DOI

2.5D Faster R-CNN for Distance Estimation

Hirotaka Hachiya, Yuki Saito, Kazuma Iteya, Masaya Nomura, Takayuki Nakamura

IEEE International Conference on Systems, Man, and Cybernetics ( IEEE )    2018.10  [Refereed]

Information-Maximization Clustering Based on Squared-Loss Mutual Information.

Masashi Sugiyama, Gang Niu, Makoto Yamada, Manabu Kimura, Hirotaka Hachiya

Neural Computation   26 ( 1 ) 84 - 131   2014  [Refereed]

DOI

Computationally Efficient Multi-Label Classification by Least-Squares Probabilistic Classifiers

Hyunha Nam, Hirotaka Hachiya, Masashi Sugiyama

IEICE TRANSACTIONS ON INFORMATION AND SYSTEMS ( IEICE-INST ELECTRONICS INFORMATION COMMUNICATIONS ENG )  E96D ( 8 ) 1871 - 1874   2013.08  [Refereed]

　View Summary

Multi-label classification allows a sample to belong to multiple classes simultaneously, which is often the case in real-world applications such as text categorization and image annotation. In multi-label scenarios, taking into account correlations among multiple labels can boost the classification accuracy. However, this makes classifier training more challenging because handling multiple labels induces a high-dimensional optimization problem. In this paper, we propose a scalable multi-label method based on the least-squares probabilistic classifier. Through experiments, we show the usefulness of our proposed method.

DOI

Feature Selection via l(1)-Penalized Squared-Loss Mutual Information

Wittawat Jitkrittum, Hirotaka Hachiya, Masashi Sugiyama

IEICE TRANSACTIONS ON INFORMATION AND SYSTEMS ( IEICE-INST ELECTRONICS INFORMATION COMMUNICATIONS ENG )  E96D ( 7 ) 1513 - 1524   2013.07  [Refereed]

　View Summary

Feature selection is a technique to screen out less important features. Many existing supervised feature selection algorithms use redundancy and relevancy as the main criteria to select features. However, feature interaction, potentially a key characteristic in real-world problems, has not received much attention. As an attempt to take feature interaction into account, we propose l(1)-LSMI, an l(1)-regularization based algorithm that maximizes a squared-loss variant of mutual information between selected features and outputs. Numerical results show that l(1)-LSMI performs well in handling redundancy, detecting non-linear dependency, and considering feature interaction.

DOI

Efficient Sample Reuse in Policy Gradients with Parameter-Based Exploration

Tingting Zhao, Hirotaka Hachiya, Voot Tangkaratt, Jun Morimoto, Masashi Sugiyama

NEURAL COMPUTATION ( MIT PRESS )  25 ( 6 ) 1512 - 1547   2013.06  [Refereed]

　View Summary

The policy gradient approach is a flexible and powerful reinforcement learning method particularly for problems with continuous actions such as robot control. A common challenge is how to reduce the variance of policy gradient estimates for reliable policy updates. In this letter, we combine the following three ideas and give a highly effective policy gradient method: (1) policy gradients with parameter-based exploration, a recently proposed policy search method with low variance of gradient estimates; (2) an importance sampling technique, which allows us to reuse previously gathered data in a consistent way; and (3) an optimal baseline, which minimizes the variance of gradient estimates with their unbiasedness being maintained. For the proposed method, we give a theoretical analysis of the variance of gradient estimates and show its usefulness through extensive experiments.

Artist Agent: A Reinforcement Learning Approach to Automatic Stroke Generation in Oriental Ink Painting

Ning Xie, Hirotaka Hachiya, Masashi Sugiyama

IEICE TRANSACTIONS ON INFORMATION AND SYSTEMS ( IEICE-INST ELECTRONICS INFORMATION COMMUNICATIONS ENG )  E96D ( 5 ) 1134 - 1144   2013.05  [Refereed]

　View Summary

Oriental ink painting, called Sumi-e, is one of the most distinctive painting styles and has attracted artists around the world. Major challenges in Sumi-e simulation are to abstract complex scene information and reproduce smooth and natural brush strokes. To automatically generate such strokes, we propose to model the brush as a reinforcement learning agent, and let the agent learn the desired brush-trajectories by maximizing the sum of rewards in the policy search framework. To achieve better performance, we provide elaborate design of actions, states, and rewards specifically tailored for a Sumi-e agent. The effectiveness of our proposed approach is demonstrated through experiments on Sumi-e simulation.

DOI

Squared-loss Mutual Information Regularization

Gang Niu, Wittawat Jitkrittum, Bo Dai, Hirotaka Hachiya, Masashi Sugiyama

International Conference on Machine Learning (ICML2013) ( International Machine Learning Society (IMLS) )    10--18   2013  [Refereed]

Relative Density-Ratio Estimation for Robust Distribution Comparison.

Makoto Yamada, Taiji Suzuki, Takafumi Kanamori, Hirotaka Hachiya, Masashi Sugiyama

Neural Computation   25 ( 5 ) 1324 - 1370   2013  [Refereed]

DOI