hh.sePublications
Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
Imbalance Data and Transfer Learning in Music Genre Classification
Halmstad University, School of Information Technology.
2025 (English)Independent thesis Basic level (degree of Bachelor), 10 credits / 15 HE creditsStudent thesis
Abstract [en]

This project investigates the performance and effects of varying degrees of data imbalanceand the role of transfer learning in music genre classification (MGC), using a relativelysmall, imbalanced music dataset labeled by domain experts. We refine the prominent deeplearning model Audio Spectrogram Transformer (AST), under these constraints. Our re-sults show that the refined model performs exceptionally well in the task of MGC, outper-forming results of previous studies. Notably, we observe that different levels of class im-balance have minimal effect on performance, contrasting previous research based on non-transformer models, which reports improvements of minority class predictions throughbalancing techniques. These findings suggest that refining a pre-trained AST model onsmall and imbalanced datasets, can still yield prominent results. Furthermore, we sug-gest that there exists indications that the AST model may be suitable as a foundationalmodel for a broader range of music and audio related tasks, both in research and practicalapplications.

Place, publisher, year, edition, pages
2025. , p. 44
Keywords [en]
Music Genre Classification, Deep Learning, Imbalance Data, Audio Spectrogram Transformer
National Category
Artificial Intelligence
Identifiers
URN: urn:nbn:se:hh:diva-56166OAI: oai:DiVA.org:hh-56166DiVA, id: diva2:1963184
Supervisors
Examiners
Available from: 2025-06-03 Created: 2025-06-02 Last updated: 2025-10-01Bibliographically approved

Open Access in DiVA

fulltext(1778 kB)163 downloads
File information
File name FULLTEXT02.pdfFile size 1778 kBChecksum SHA-512
0c76e2e48bde2ca40085cd5b7f9ea16649186b42dc0565e53c52cd6cf400a0528629c86d622d680f88ae252e6a6c281e497fd7dcbbd8d946c6aac552658db155
Type fulltextMimetype application/pdf

By organisation
School of Information Technology
Artificial Intelligence

Search outside of DiVA

GoogleGoogle Scholar
Total: 163 downloads
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

urn-nbn

Altmetric score

urn-nbn
Total: 294 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf