Published Papers - HACHIYA Hirotaka

FIRE-AD: frequency-dependent image reconstruction error for micro defect detection

Yuhei Nomura, Hirotaka Hachiya （Part： Last author )

Machine Vision and Applications     2025.07  [Refereed]

DOI

Enhancing semantic audio-visual representation learning with supervised multi-scale attention

Jiwei Zhang, Yi Yu, Suhua Tang, GuoJun Qi, Haiyuan Wu, Hirotaka Hachiya （Part： Last author )

Pattern Analysis and Applications ( Springer Science and Business Media LLC )  28 ( 2 )   2025.02  [Refereed]

DOI

Deep Supervised with Fine-grained Feature Fusion Network for Cross-modal Retrieval

Jiwei Zhang, Hirotaka Hachiya （Part： Last author )

International Conference on Artificial Intelligence in Information and Communication (ICAIIC)     2025.02  [Refereed]

Randomized Channel-Pass Mask for Channel-Wise Explanation of Black-Box Models

Hirotaka Hachiya, Daiki Nisawa （Part： Lead author )

Lecture Notes in Computer Science ( Springer Nature Singapore )    454 - 468   2024.12

DOI

MLP-Mixer based surrogate model for seismic ground motion with spatial source and geometry parameters

Hirotaka Hachiya, Yuto Kuroki, Asako Iwaki, Takahiro Maeda, Naonori Ueda, Hiroyuki Fujiwara （Part： Lead author )

Proceedings of Asian Conference on Machine Learning (ACML)     2024.12  [Refereed]

Specular Surface Detection with Deep Static Specular Flow and Highlight

Hirotaka Hachiya, Yuto Yoshimura （Part： Lead author )

Machine Vision and Applications ( Springer Science and Business Media LLC )  35 ( 6 )   2024.09  [Refereed]

DOI

Permutation Dependent Feature Mixing in TSMixer for Multivariate Time Series Forecasting

Rikuto Yamazono, Hirotaka Hachiya （Part： Last author )

Proceedings of European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases (ECMLPKDD), 2024     2024.09  [Refereed]

Multi-class AUC maximization for imbalanced ordinal multi-stage tropical cyclone intensity change forecast

Hirotaka Hachiya, Hiroki Yoshida, Udai Shimada, Naonori Ueda （Part： Lead author )

Machine Learning with Applications ( Elsevier BV )  17   100569 - 100569   2024.09  [Refereed]

DOI

A Multi‐Fusion Residual Attention U‐Net Using Temporal Information for Segmentation of Left Ventricular Structures in 2D Echocardiographic Videos

Kai Wang, Hirotaka Hachiya, Haiyuan Wu （Part： Last author )

International Journal of Imaging Systems and Technology ( Wiley )  34 ( 4 )   2024.07  [Refereed]

　View Summary

ABSTRACT

The interpretation of cardiac function using echocardiography requires a high level of diagnostic proficiency and years of experience. This study proposes a multi‐fusion residual attention U‐Net, MURAU‐Net, to construct automatic segmentation for evaluating cardiac function from echocardiographic video. MURAU‐Net has two benefits: (1) Multi‐fusion network to strengthen the links between spatial features. (2) Inter‐frame links can be established to augment the temporal coherence of sequential image data, thereby enhancing its continuity. To evaluate the effectiveness of the proposed method, we performed nine‐fold cross‐validation using CAMUS dataset. Among state‐of‐the‐art methods, MURAU‐Net achieves highly competitive score, for example, Dice similarity of 0.952 (ED phase) and 0.931 (ES phase) in , 0.966 (ED phase) and 0.957 (ES phase) in , and 0.901 (ED phase) and 0.917 (ES phase) in , respectively. It also achieved the Dice similarity of 0.9313 in the EchoNet‐Dynamic dataset for the overall left ventricle segmentation. In addition, we show MURAU‐Net can accurately segment multiclass cardiac ultrasound videos and output the animation of segmentation results using the original two‐chamber cardiac ultrasound dataset MUCO.

DOI

Set representative vector and its asymmetric attention-based transformation for heterogeneous set-to-set matching

Hirotaka Hachiya, Yuki Saito （Part： Lead author )

Neurocomputing     2024.04  [Refereed]

DOI

A Transformer-based fully trainable point process

Hirotaka HACHIYA, Fumiya NISHIZAWA

IEICE Transactions on Information and Systems ( Institute of Electronics, Information and Communications Engineers (IEICE) )    2024  [Refereed]

DOI

INTERPRETABLE DEEP INPAINTING BASED ON SUBSURFACE STRUCTURE DATA FOR SPATIAL INTERPOLATION OF SEISMIC MOTIONS

HACHIYA Hirotaka, TARASUKI Yuka, IWAKI Asako, MAEDA Takahiro, UEDA Naonori, FUJIWARA Hiroyuki （Part： Lead author )

Journal of Japan Association for Earthquake Engineering ( Japan Association for Earthquake Engineering )  24 ( 5 ) 5_35 - 5_44   2024  [Refereed]

DOI

Frequency-Dependent Image Reconstruction Error for Micro Defect Detection

Yuhei Nomura, Hirotaka Hachiya （Part： Last author )

Proceedings of The 15th Asian Conference on Machine Learning (ACML2023)     2023.11  [Refereed]

Position-dependent partial convolutions for supervised spatial interpolation

Hirotaka Hachiya, Kotaro Nagayoshi, Asako Iwaki, Takahiro Maeda, Naonori Ueda, Hiroyuki Fujiwara （Part： Lead author )

Machine Learning with Applications ( Elsevier BV )    100514 - 100514   2023.11  [Refereed]

DOI

Combining Static Specular Flow and Highlight with Deep Features for Specular Surface Detection

Hirotaka Hachiya, Yuta Yoshimura （Part： Lead author )

Proceedings of The 18th International Conference on Machine Vision Applications (MVA2023)     2023.07  [Refereed]

Encoder–decoder-based image transformation approach for integrating multiple spatial forecasts

Hirotaka Hachiya, Yusuke Masumoto, Atsushi Kudo, Naonori Ueda （Part： Lead author )

Machine Learning with Applications ( Elsevier BV )  12 ( 100473 ) 1 - 11   2023.05  [Refereed]

DOI

Multistream-Based Marked Point Process With Decomposed Cumulative Hazard Functions

Hirotaka Hachiya, Sujun Hong （Part： Lead author )

Neural Computation ( MIT Press )  35 ( 4 ) 699 - 726   2023.03  [Refereed]

　View Summary

Abstract

When applying a point process to a real-world problem, an appropriate intensity function model should be designed based on physical and mathematical prior knowledge. Recently, a fully trainable deep learning–based approach has been developed for temporal point processes. In this approach, a cumulative hazard function (CHF) capable of systematic computation of adaptive intensity function is modeled in a data-driven manner. However, in this approach, although many applications of point processes generate various kinds of information such as location, magnitude, and depth, the mark information of events is not considered. To overcome this limitation, we propose a fully trainable marked point process method for modeling decomposed CHFs for time and mark prediction using multistream deep neural networks. We demonstrate the effectiveness of the proposed method through experiments with synthetic and real-world event data.

DOI

Multi‐feature subspace representation network for person re‐identification via bird's‐eye view image

Jiwei Zhang, Haiyuan Wu, Qian Chen, Hirotaka Hachiya

Computer Animation and Virtual Worlds ( Wiley )    2023.02  [Refereed]

DOI

Position-dependent partial convolutions for supervised spatial interpolation

Hirotaka Hachiya, Kotaro Nagayoshi, Asako Iwaki, Takahiro Maeda, Naonori Ueda, Hiroyuki Fujiwara （Part： Lead author )

Proceedings of The 14th Asian Conference on Machine Learning (ACML2022)   189   420 - 435   2022.12  [Refereed]

Study on Echocardiographic Image Segmentation Based on Attention U-Net

Kai Wang, Jiwei Zhang, Hirotaka Hachiya, Haiyuan Wu

2022 IEEE International Conference on Mechatronics and Automation (ICMA) ( IEEE )    2022.08  [Refereed]

DOI