HTM Spatial Pooler - a Nonparametric Interpretable Feature Selection Algorithm? An Introductory Exploration
Neuroimaging data analysis suffers from the curse of dimensionality. Feature selection plays an important role in mitigating this curse. Here we explore the potential of the Hierarchical Temporal Memory (HTM) Spatial Pooler (SP) as a nonparametric interpretable feature selection algorithm for high-dimensional neuroimaging data. Our results indicate that SP demonstrates comparable performance to other feature selection algorithms in terms of feature stability and classification performance.
Authors
- Drozdstoy Stoyanov (ORCID: https://orcid.org/0000-0002-9975-3680)
- Rositsa Paunova (ORCID: https://orcid.org/0000-0002-9592-645X)
- Alexander E. Hramov (ORCID: https://orcid.org/0000-0003-2787-2530)
- Semen Kurkin (ORCID: https://orcid.org/0000-0002-3438-5717)
- Muhammad Salman Kabir (ORCID: https://orcid.org/0000-0002-0071-6901)
Institutions
- Immanuel Kant Baltic Federal University (RU)
- Medical University Plovdiv (BG)
- Innopolis University (RU)
Publication Details
- Published
- 2024-09-19
- DOI
- https://doi.org/10.1109/dcna63495.2024.10718624
- Primary Topic
- Machine Learning and Data Classification
- Type
- article
- Field-Weighted Citation Impact
- 0.00
Funders
- Russian Science Foundation