Updated on 2025/05/02

写真a

 
Daiki Koge
 
Organization
Academic Assembly Institute of Science and Technology JOUHOU DENSHI KOUGAKU KEIRETU Assistant Professor
Graduate School of Science and Technology Electrical and Information Engineering Assistant Professor
Faculty of Engineering Department of Engineering Assistant Professor
Title
Assistant Professor
Contact information
メールアドレス
Homepage
https://sites.google.com/view/daiki-ko/%E3%83%9B%E3%83%BC%E3%83%A0/
Other name(s)
Daiki Koge
▶ display Profile on researchmap
External link

Degree

  • 博士（工学） ( 2024.3   奈良先端科学技術大学院大学 )

  • 修士（工学） ( 2021.3   奈良先端科学技術大学院大学 )

  • 学士（工学） ( 2019.3   鹿児島大学 )

Research Interests

  • Bayesian Inference

  • Cheminformatics

  • Probabilistic model

  • Deep Learning

  • Machine Learning

  • データ駆動型化学

Research Areas

  • Informatics / Life, health and medical informatics  / Cheminformatics

Research History (researchmap)

  • 新潟大学大学院 自然科学研究科   助教

    2024.4

    　 More details

  • 新潟大学 工学部   助教

    2024.4

    　 More details

  • Japan Society for the Promotion of Science   JSPS DC2

    2022.4 - 2024.3

    　 More details

  • Nara Institute of Science and Technology   先端科学技術研究科   PhD Student

    2021.4 - 2024.3

    　 More details

Research History

  • Niigata University   Electrical and Information Engineering, Graduate School of Science and Technology   Assistant Professor

    2024.4

  • Niigata University   Department of Engineering, Faculty of Engineering   Assistant Professor

    2024.4

  • Niigata University   Institute of Science and Technology, Academic Assembly   Assistant Professor

    2024.4

 

Papers

  • Generalized Molecular Latent Representation via Graph Latent Diffusion Autoencoder

    Daiki Koge, Naoaki Ono, Takashi Abe, Shigehiko Kanaya

    2025.3

    　More details

    Authorship：Lead author   Publisher：Springer Science and Business Media LLC  

    Abstract <p>In recent years, deep neural networks (DNNs) have been applied for constructing molecular latent representations for drug discovery. The quality of these representations, obtained using DNN encoders, affects the generalization performance of the model, i.e., its ability to predict molecular properties for previously unseen compounds. Given the vast space of potential organic compounds and limited availability of labeled data with specific molecular properties, enhancing the generalization performance of predictive models is key for accelerating drug discovery. This requires the construction of effective molecular latent representations. Considering this aspect, this paper introduces the graph latent diffusion autoencoder (Graph LDA), a deep molecular generative model that combines a graph-transformer-based variational autoencoder and latent-diffusion-based latent prior model, designed for constructing generalized molecular representation through unsupervised learning. To assess the generalization performance of molecular property predictions based on the constructed molecular representations, the results for Graph LDA were compared with those of existing models using the widely applicable information criterion (WAIC) and widely applicable Bayesian information criterion (WBIC). The results indicated that Graph LDA outperformed the existing methods. Furthermore, we empirically demonstrated that the superior generalization performance of Graph LDA is attributable to the smoothness and multimodality of its learned molecular latent representation. The proposed robust framework for molecular property prediction holds significant potential for accelerating drug discovery and material advancements.Scientific contribution :This work introduces Graph LDA, a novel deep molecular generative model that combines a graph-transformer-based variational autoencoder and latent-diffusion-based latent prior model. The proposed model can extract smooth molecular latent representations with multimodal distributions, resulting in high generalization performance for molecular property prediction. Results of WAIC and WBIC analyses demonstrate that Graph LDA significantly outperforms existing representative unsupervised representation learning models.</p>

    DOI： 10.21203/rs.3.rs-6299941/v1

    researchmap

    Other Link： https://www.researchsquare.com/article/rs-6299941/v1.html

  • Pre-training of Molecular GNNs via Conditional Boltzmann Generator

    Daiki Koge, Naoaki Ono, Shigehiko Kanaya

    2023.12

    　More details

    Authorship：Lead author  

    File： 2312.13110.pdf

    researchmap

  • Variational Autoencoding Molecular Graphs with Denoising Diffusion Probabilistic Model

    Daiki Koge, Naoaki Ono, Shigehiko Kanaya

    arXiv   2023.7

    　More details

    Authorship：Lead author  

    File： GraphLDA.pdf

    researchmap

  • Classification of metabolites by metabolic pathways concerning terpenoids, phenylpropanoids, and polyketide compounds based on machine learning Reviewed

    Yuri Koide, Daiki Koge, Shigehiko Kanaya, Md. Altaf-Ul-Amin, Ming Huang, Aki Hirai Morita, Naoaki Ono

    Journal of Computer Aided Chemistry   23   25 - 34   2023

    　More details

    Publishing type：Research paper (scientific journal)   Publisher：Division of Chemical Information and Computer Sciences  

    DOI： 10.2751/jcac.23.25

    researchmap

  • Embedding of Molecular Structure Using Molecular Hypergraph Variational Autoencoder with Metric Learning Reviewed

    Daiki Koge, Naoaki Ono, Ming Huang, Md. Altaf‐Ul‐Amin, Shigehiko Kanaya

    Molecular Informatics   40 ( 2 )   2000203 - 2000203   2020.11

    　More details

    Authorship：Lead author   Publishing type：Research paper (scientific journal)   Publisher：Wiley  

    DOI： 10.1002/minf.202000203

    researchmap

    Other Link： https://onlinelibrary.wiley.com/doi/full-xml/10.1002/minf.202000203

  • Data Science for QSAR for Protease activity Reviewed

    Hideki Ueda, Akio Fukumori, Daiki Koge, Naoaki Ono, Md. Altaf-Ul-Amin, Shigehiko Kanaya

    Journal of Computer Aided Chemistry   23   43 - 49   2023

    　More details

    Publishing type：Research paper (scientific journal)   Publisher：Division of Chemical Information and Computer Sciences  

    DOI： 10.2751/jcac.23.43

    researchmap

  • Molecular Graph Indexes for Assessing Heterogeneity of Chemical Compounds Reviewed

    Keisuke Wakakuri, Yudai Taguchi, Daiki Koge, Naoaki Ono, Md. Altaf-Ul-Amin, Shigehiko Kanaya

    Journal of Computer Aided Chemistry   23   50 - 59   2023

    　More details

    Publishing type：Research paper (scientific journal)   Publisher：Division of Chemical Information and Computer Sciences  

    DOI： 10.2751/jcac.23.50

    researchmap

▶ display all

Books

  • データ駆動型材料開発 : オントロジーとマイニング、計測と実験装置の自動制御

    （ Role： Contributor ,  第４章：構造・物性探索と機械学習）

    エヌ・ティー・エス  2021.11  （ ISBN:9784860437596 ）

    　More details

    Total pages：3, 6, 244, 6, 図版26p   Language：Japanese

    CiNii Books

    researchmap

Presentations

  • Generalized Molecular Representation for Drug Discovery via Molecular Graph Latent Diffusion Autoencoder

    Daiki Koge

    Chem-Bio Informatics Society (CBI) 2024  2024.10 

    　More details

    Event date： 2024.10

    researchmap

  • Variational Autoencoding Molecular Graphs with Denoising Diffusion Probabilistic Model

    Daiki Koge

    IEEE CIBCB 2023  2023.8 

    　More details

Awards

  • Molecular Informatics : Top cited paper among work pulished in an issue between 1 January 2021 - 15 December 2022

    2023.4   Embedding of Molecular Structure Using Molecular Hypergraph Variational Autoencoder with Metric Learning

    Daiki Koge, Naoaki Ono, Ming Huang, Md. Altaf-Ul-Amin, Shigehiko Kanaya

    　More details

  • 卒業論文優秀発表賞

    2019.3   鹿児島大学 工学部   ひまわり8号の画像解析による桜島噴火後の火山灰追尾

    高下大貴

    　More details

Research Projects

  • 深層学習による分子立体構造の動的揺らぎに関する特徴量の抽出と分子特性予測への応用

    Grant number：24K23886

    2024.7 - 2026.3

    System name：科学研究費助成事業

    Research category：研究活動スタート支援

    Awarding organization：日本学術振興会

    高下 大貴

    　 More details

    Grant amount：\2080000 （ Direct Cost: \1600000 、 Indirect Cost：\480000 ）

    researchmap

  • 深層学習を用いた分子埋め込みモデルによる薬剤候補分子の仮想的探索

    Grant number：22KJ2285

    2022.4 - 2024.3

    System name：科学研究費助成事業

    Research category：特別研究員奨励費

    Awarding organization：日本学術振興会

    高下 大貴, 高下 大貴

    　 More details

    Grant amount：\1700000 （ Direct Cost: \1700000 ）

    researchmap

Other research activities

  • 令和３年度 奈良先端科学技術大学院大学科学技術イノベーション創出に向けた大学フェローシップ

    2021.4
    -
    2022.3

    　More details

  • 日本学生支援機構奨学金 業績優秀者 返還免除 (全額)

    　More details

    修士課程分

    researchmap

 

Teaching Experience

  • 研究室体験実習

    2024
    Institution name：新潟大学

  • 情報システム基礎実習

    2024
    Institution name：新潟大学

  • 人工知能特論

    2024
    Institution name：新潟大学