*Result*: Activity Recognition from Daily-Life Sounds Using Unsupervised Learning with Dirichlet Multinomial Mixture Models.

Title:
Activity Recognition from Daily-Life Sounds Using Unsupervised Learning with Dirichlet Multinomial Mixture Models.
Authors:
Sadohara K; National Institute of Advanced Industrial Science and Technology (AIST), 2-3-26 Aomi, Koto-ku, Tokyo 135-0064, Japan., Miyata N; National Institute of Advanced Industrial Science and Technology (AIST), 2-3-26 Aomi, Koto-ku, Tokyo 135-0064, Japan.
Source:
Sensors (Basel, Switzerland) [Sensors (Basel)] 2026 Feb 27; Vol. 26 (5). Date of Electronic Publication: 2026 Feb 27.
Publication Type:
Journal Article
Language:
English
Journal Info:
Publisher: MDPI Country of Publication: Switzerland NLM ID: 101204366 Publication Model: Electronic Cited Medium: Internet ISSN: 1424-8220 (Electronic) Linking ISSN: 14248220 NLM ISO Abbreviation: Sensors (Basel) Subsets: MEDLINE
Imprint Name(s):
Original Publication: Basel, Switzerland : MDPI, c2000-
Grant Information:
JPNP20006 New Energy and Industrial Technology Development Organization
Contributed Indexing:
Keywords: Dirichlet multinomial mixture model (DMM); acoustic scene classification; activity recognition; ambient assisted living; bespoke system; burstiness; neural audio codec; topic model
Entry Date(s):
Date Created: 20260314 Date Completed: 20260314 Latest Revision: 20260316
Update Code:
20260316
PubMed Central ID:
PMC12987356
DOI:
10.3390/s26051509
PMID:
41829470
Database:
MEDLINE

*Further Information*

*To support ambient assisted living for the elderly living alone, we investigate a method for recognizing daily activities from household sounds. To reduce the cost of building an activity-recognition model, we adopt an unsupervised learning approach based on a Dirichlet multinomial mixture model. The model represents the generative process of neural audio codec codes conditioned on latent activities. We further extend the model to handle multiple streams of codes corresponding to different sound directions. This extension enables the formation of more accurate activity clusters, partly because code occurrence patterns exhibit burstiness. The proposed approach is expected to serve as a key component for constructing an activity recognition system that requires minimal labeled data and a small number of user inquiries.*