• Search Research Projects
  • Search Researchers
  • How to Use
  1. Back to previous page

Construction of estimation methods and their uncertainty quantification methods for high-dimensional count data focusing on structural constraints

Research Project

Project/Area Number 19K20222
Research Category

Grant-in-Aid for Early-Career Scientists

Allocation TypeMulti-year Fund
Review Section Basic Section 60030:Statistical science-related
Research InstitutionThe Institute of Statistical Mathematics (2020-2023)
The University of Tokyo (2019)

Principal Investigator

Yano Keisuke  統計数理研究所, 統計基盤数理研究系, 准教授 (20806070)

Project Period (FY) 2019-04-01 – 2024-03-31
Project Status Completed (Fiscal Year 2023)
Budget Amount *help
¥4,030,000 (Direct Cost: ¥3,100,000、Indirect Cost: ¥930,000)
Fiscal Year 2022: ¥1,040,000 (Direct Cost: ¥800,000、Indirect Cost: ¥240,000)
Fiscal Year 2021: ¥780,000 (Direct Cost: ¥600,000、Indirect Cost: ¥180,000)
Fiscal Year 2020: ¥780,000 (Direct Cost: ¥600,000、Indirect Cost: ¥180,000)
Fiscal Year 2019: ¥1,430,000 (Direct Cost: ¥1,100,000、Indirect Cost: ¥330,000)
Keywords情報量規準 / ベイズ予測 / MCMC / 予測分布 / ベイズ統計 / Bayesian prediction / Hierarchical model / Markov chain Monte Carlo / divergence / information criterion / 不均一性 / 類似度学習 / リンク回帰 / 高次元統計 / カウントデータ / スパースモデリング / 擬ベイズ法
Outline of Research at the Start

高次元カウントデータに関する高精度かつ高速な統計解析手法を構築する。高次元カウントデータは多岐にわたる学術分野や実社会で現れる。例えば、商品の購入者数、犯罪発生件数、地震の発生件数、遺伝子の発現数、太陽の黒点数などである。近年、擬疎性という実データに現れる構造制約に着目して高精度な未知母数推定法を構築されつつある。より幅広い高次元カウントデータに適用するためには「擬疎性以外の構造制約の考慮」が必要であり推定結果の公開には「実用的な不確実性評価法の構築」が不可欠である。本研究では「構造制約に着目した高次元カウントデータの未知母数推定法と不確実性評価法の構築」を目指す。

Outline of Final Research Achievements

This project has established evaluation methods for predictive models based on Bayesian predictive distributions that are applicable to high-dimensional models including count data models. The Widely Applicable Information Criterion (WAIC) has been extensively used for evaluating predictive models based on Bayesian predictive distributions. We demonstrated the theoretical validity of WAIC in high-dimensional models and established efficient computational methods for high-dimensional models, including deep learning. Furthermore, we established an extension of WAIC, the Posterior Covariance Information Criterion (PCIC), which accommodates cases where there are weights on observations, different evaluation functions for predictions and observations, and predictive evaluation functions other than the logarithmic loss.

Academic Significance and Societal Importance of the Research Achievements

高次元モデルやカウントデータモデルは諸科学で広く現れる統計モデルである。しかし、その推論法は通常のモデルと比べて十分に定まっているとはいえない。本研究では高次元モデルやカウントデータモデルで利用可能なベイズ予測分布に基づく予測モデルの評価法を確立した。これにより従来はできなかった予測評価(深層学習を含む高次元モデルでの予測評価・観測の重みが存在する場合の予測評価・予測と観測の評価関数が異なる場合の予測評価・対数損失以外の予測評価関数を用いた場合の評価)が可能となった。

Report

(6 results)
  • 2023 Annual Research Report   Final Research Report ( PDF )
  • 2022 Research-status Report
  • 2021 Research-status Report
  • 2020 Research-status Report
  • 2019 Research-status Report
  • Research Products

    (25 results)

All 2024 2023 2022 2021 2020 2019 Other

All Int'l Joint Research (1 results) Journal Article (8 results) (of which Int'l Joint Research: 2 results,  Peer Reviewed: 8 results,  Open Access: 4 results) Presentation (16 results) (of which Int'l Joint Research: 8 results,  Invited: 14 results)

  • [Int'l Joint Research] University of Southern California/Data Sciences and Operations(米国)

    • Related Report
      2019 Research-status Report
  • [Journal Article] Minimum information dependence modeling2024

    • Author(s)
      Tomonari Sei and Keisuke Yano
    • Journal Title

      Bernoulli

      Volume: -

    • Related Report
      2023 Annual Research Report
    • Peer Reviewed
  • [Journal Article] Posterior Covariance Information Criterion for Weighted Inference.2023

    • Author(s)
      Yukito Iba, Keisuke Yano
    • Journal Title

      Neural Computation

      Volume: 35 Issue: 7 Pages: 1340-1361

    • DOI

      10.1162/neco_a_01592

    • Related Report
      2023 Annual Research Report 2022 Research-status Report
    • Peer Reviewed
  • [Journal Article] A generalization gap estimation for overparameterized models via the Langevin functional variance2023

    • Author(s)
      Okuno Akifumi、Yano Keisuke
    • Journal Title

      Journal of Computational and Graphical Statistics

      Volume: 1 Issue: 4 Pages: 1-20

    • DOI

      10.1080/10618600.2023.2197488

    • Related Report
      2023 Annual Research Report 2022 Research-status Report
    • Peer Reviewed
  • [Journal Article] Dependence of variance on covariate design in nonparametric link regression2023

    • Author(s)
      Okuno Akifumi、Yano Keisuke
    • Journal Title

      Statistics & Probability Letters

      Volume: 193 Pages: 109716-109716

    • DOI

      10.1016/j.spl.2022.109716

    • Related Report
      2022 Research-status Report
    • Peer Reviewed / Open Access
  • [Journal Article] Optimal Shrinkage Estimation of Predictive Densities Under α-Divergences2021

    • Author(s)
      George Edward、Mukherjee Gourab、Yano Keisuke
    • Journal Title

      Bayesian Analysis

      Volume: 16 Issue: 4 Pages: 1139-1155

    • DOI

      10.1214/21-ba1264

    • Related Report
      2021 Research-status Report 2020 Research-status Report
    • Peer Reviewed / Open Access / Int'l Joint Research
  • [Journal Article] Minimax predictive density for sparse count data2021

    • Author(s)
      Yano Keisuke、Kaneko Ryoya、Komaki Fumiyasu
    • Journal Title

      Bernoulli

      Volume: 27 Issue: 2 Pages: 1212-1238

    • DOI

      10.3150/20-bej1271

    • Related Report
      2021 Research-status Report 2020 Research-status Report
    • Peer Reviewed / Open Access
  • [Journal Article] On frequentist coverage errors of Bayesian credible sets in moderately high dimensions2020

    • Author(s)
      Keisuke Yano and Kengo Kato
    • Journal Title

      Bernoulli

      Volume: 26 Issue: 1 Pages: 616-641

    • DOI

      10.3150/19-bej1142

    • Related Report
      2019 Research-status Report
    • Peer Reviewed / Int'l Joint Research
  • [Journal Article] Adjacency-based regularization for partially ranked data with non-ignorable missing2020

    • Author(s)
      Kento Nakamura, Keisuke Yano, and Fumiyasu Komaki
    • Journal Title

      Computational Statistics & Data Analysis

      Volume: 145 Pages: 106905-106905

    • DOI

      10.1016/j.csda.2019.106905

    • Related Report
      2019 Research-status Report
    • Peer Reviewed / Open Access
  • [Presentation] A new approach to mixed-domain and higher-order dependence modeling2023

    • Author(s)
      Keisuke Yano
    • Organizer
      Global Plasma Forum in Aomori
    • Related Report
      2023 Annual Research Report
    • Int'l Joint Research / Invited
  • [Presentation] 最小情報従属モデルを用いた混合ドメイン多変量解析2023

    • Author(s)
      Keisuke Yano
    • Organizer
      令和5年度第2回 日本大学生産工学部人工知能リサーチセンター講演会
    • Related Report
      2023 Annual Research Report
    • Invited
  • [Presentation] 重み付き推論における汎化性能推定のための事後共分散型情報量規準2023

    • Author(s)
      Keisuke Yano
    • Organizer
      大阪大学 数理・データ科学セミナー データ科学セミナーシリーズ
    • Related Report
      2023 Annual Research Report
    • Invited
  • [Presentation] 高次元・無限次元モデルにおける予測分布2022

    • Author(s)
      矢野 恵佑
    • Organizer
      日本統計学会各賞受賞者記念講演
    • Related Report
      2022 Research-status Report
    • Invited
  • [Presentation] 予測分布論の最近の進展2021

    • Author(s)
      矢野 恵佑
    • Organizer
      日本数学会
    • Related Report
      2021 Research-status Report
    • Invited
  • [Presentation] On estimating generalization gaps via the functional variance in overparameterized models2021

    • Author(s)
      Keisuke Yano
    • Organizer
      CMStatistics2021
    • Related Report
      2021 Research-status Report
    • Invited
  • [Presentation] 予測の情報量規準2021

    • Author(s)
      矢野 恵佑
    • Organizer
      情報計測オンラインセミナー
    • Related Report
      2021 Research-status Report
    • Invited
  • [Presentation] 擬ベイズ事後分布に基づく予測評価のための情報量規準2021

    • Author(s)
      矢野恵佑, 伊庭幸人
    • Organizer
      2021年度統計関連学会連合大会
    • Related Report
      2021 Research-status Report
  • [Presentation] Risk-estimation based predictive densities for heteroskedastic hierarchical models2019

    • Author(s)
      Keisuke Yano
    • Organizer
      ICSA 2019, China
    • Related Report
      2019 Research-status Report
    • Int'l Joint Research / Invited
  • [Presentation] On estimation and prediction for high-dimensional Poisson models with quasi-zero inflation2019

    • Author(s)
      Keisuke Yano
    • Organizer
      CMStatistics 2019 , UK
    • Related Report
      2019 Research-status Report
    • Int'l Joint Research / Invited
  • [Presentation] On frequentist coverage errors of Bayesian credible sets in moderately high dimensions, Italy, 24 July (22-26 July), 2019.2019

    • Author(s)
      Keisuke Yano
    • Organizer
      European Meeting of Statisticians (EMS 2019), Italy
    • Related Report
      2019 Research-status Report
    • Int'l Joint Research / Invited
  • [Presentation] The Berry--Esseen type bound for the Bernstein--von Mises theorem in moderately high dimensions2019

    • Author(s)
      Keisuke Yano
    • Organizer
      EAC-ISBA 2019, Japan
    • Related Report
      2019 Research-status Report
    • Int'l Joint Research / Invited
  • [Presentation] On the construction of adaptive predictive densities for sparse count data2019

    • Author(s)
      Keisuke Yano
    • Organizer
      EcoSta 2019, Taiwan
    • Related Report
      2019 Research-status Report
    • Int'l Joint Research / Invited
  • [Presentation] Adaptive minimax predictive density for sparse Poisson models2019

    • Author(s)
      Keisuke Yano
    • Organizer
      Banff workshop "New and Evolving Roles of Shrinkage in Large-Scale Prediction and Inference (19w5188)", Canada
    • Related Report
      2019 Research-status Report
    • Int'l Joint Research / Invited
  • [Presentation] Earthquake detection using deep learning for continuous seismic network records2019

    • Author(s)
      Keisuke Yano, Takahiro Shiina, Sumito Kurata, Aitaro Kato, Fumiyasu Komaki, Shin'ichi Sakai, Naoshi Hirata
    • Organizer
      StatSei11, Japan
    • Related Report
      2019 Research-status Report
    • Int'l Joint Research / Invited
  • [Presentation] ベイズl1トレンドフィルタリングによるスロースリップ自動検知2019

    • Author(s)
      矢野恵佑, 加納将行
    • Organizer
      固体地球科学データ同化に関する研究会
    • Related Report
      2019 Research-status Report

URL: 

Published: 2019-04-18   Modified: 2025-01-30  

Information User Guide FAQ News Terms of Use Attribution of KAKENHI

Powered by NII kakenhi