Hierarchical Topic Model for Tensor Data and Extraction of Weekly and Daily Patterns from Activity Monitor Records

Shunichi Nomura, Michiko Watanabe, Yuko Oguma

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Latent Dirichlet allocation (LDA) is a popular topic model for extracting common patterns from discrete datasets. It is extended to the pachinko allocation model (PAM) with a hierarchical topic structure. This paper presents a combination meal allocation (CMA) model, which is a further enhanced topic model from the PAM that has both hierarchical categories and hierarchical topics. We consider count datasets in multiway arrays, i.e., tensors, and introduce a set of topics to each mode of the tensors. The topics in each mode are interpreted as patterns in the topics and categories in the next mode. Despite there being a vast number of combinations in multilevel categories, our model provides simple and interpretable patterns by sharing the topics in each mode. Latent topics and their membership are estimated using Markov chain Monte Carlo (MCMC) methods. We apply the proposed model to step-count data recorded by activity monitors to extract some common activity patterns exhibited by the users. Our model identifies four daily patterns of ambulatory activities (commuting, daytime, nighttime, and early-bird activities) as sub-topics, and six weekly activity patterns as super-topics. We also investigate how the amount of activity in each pattern dynamically affects body weight changes.

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume12705 LNAI
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

ConferenceWorkshop on Smart and Precise Agriculture, WSPA 2021, PAKDD 2021 Workshop on Machine Learning for MEasurement Informatics, MLMEIN 2021, 1st Workshop and Shared Task on Scope Detection of the Peer Review Articles, SDPRA 2021, 1st International Workshop on Data Assessment and Readiness for AI, DARAI 2021 and 1st International Workshop on Artificial Intelligence for Enterprise Process Transformation, AI4EPT 2021 held in conjunction with 25th Pacific-Asia Conference on Knowledge Discovery and Data Mining, PAKDD 2021
CityVirtual, Online
Period21/5/1121/5/14

Keywords

  • Activity monitor records
  • Tensor data
  • Topic model

ASJC Scopus subject areas

  • Theoretical Computer Science
  • General Computer Science

Fingerprint

Dive into the research topics of 'Hierarchical Topic Model for Tensor Data and Extraction of Weekly and Daily Patterns from Activity Monitor Records'. Together they form a unique fingerprint.

Cite this