Evidence record · UDINUS · SDGs
- Unit
- Rektorat UDINUS
- Period
- 2023
- Goals
- Record
- dbcb3c31-cbf6-477b-83db-33c0098a197c
- Published
- 2026-10-30
Universitas Dian Nuswantoro conducted a study on improving birdsong classification through the selection of the most pertinent features. The activity involved UDINUS researchers Pramunendar, Ricardus Anggi; Andono, Pulung Nurtantio; Shidik, Guruh Fajar; Megantara, Rama Aria; Pergiwati, Dewi; Prabowo, Dwi Puji; Sari, Yuslena; and Soong, Lim Way. The study addressed the need for more precise classification because birds are described as exceptional environmental regulators and many species are endangered. More precise classification can assist the community in distinguishing bird species and conserving the local environment.
The workflow addressed disruptive noise, unfavorable qualities in birdsong, and the constraint-management challenges associated with high-dimensional data. Outliers were removed through the application of k-means clustering. Noise was reduced using YAMNet, while feature synthesis was performed using gammatone cepstral coefficients, identified in the source as GFCC. After these stages, the Grey Wolf Optimizer was employed to identify the most optimum features from the data. The fitness functions used in the research were derived from the K-nearest neighbors algorithm. The study therefore focused on using GWO and KNN within a workflow for managing high-dimensional birdsong-classification data.
According to the evaluation outcomes, the number of features was reduced by more than 30.7% compared with the initial number of features, while the resulting accuracy was 81.06%. The source states that this discovery improved precision by 4% and surpassed prior research. The work presents the optimization method, particularly GWO, as effective in the reported classification setting and describes its contribution as a new workflow for analyzing high-dimensional data and enhancing birdsong classification. The activity is connected to the stated purpose of supporting conservation of the local environment through its focus on birds and species distinction. The study frames UDINUS's contribution through computational research addressing birdsong analysis and the management of data constraints.