Sethi, Deeksha and Nagaraj, Nithin and Harikrishnan, Nellippallil Balakrishnan (2023) Neurochaos feature transformation for Machine Learning. Integration, 90. pp. 157-162.
![[img]](http://eprints.nias.res.in/style/images/fileicons/text.png) |
Text
2023-Neurochaos_feature_transformation_for_Machine_Learning_NithinNagaraj.pdf - Published Version Restricted to Registered users only Download (1MB) |
| Contribution | Name |
|---|
| Abstract: | With today’s increasing data complexity, efficient feature extraction has become an integral part of learning. In this spirit, we use features from a recently proposed brain-inspired learning algorithm for classification namely Neurochaos Learning (NL). This paper compares NL: chaos-based hybrid ML architectures to stand-alone ML/NL architectures. The maximum performance boost obtained is 25.97% (Statlog (Heart) dataset using NL features+Decision Tree) in the high training sample regime and 144.38% (Haberman’s Survival dataset using NL features+Random Forest) in the low training sample regime. NL offers enormous flexibility in integrating Neurochaos features with any ML classifier. |
|---|---|
| Item Type: | Journal Paper |
| Subjects: | School of Humanities > Consciousness Studies School of Humanities > Cognitive Science |
| Divisions: | Schools > Humanities |
| Date Deposited: | 11 Apr 2023 04:16 |
| Last Modified: | 11 Apr 2023 04:16 |
| Official URL: | https://www.sciencedirect.com/science/article/abs/... |
| Related URLs: | |
| Funders: | * |
| Projects: | * |
| DOI: | https://doi.org/10.1016/j.vlsi.2023.01.014 |
| URI: | http://eprints.nias.res.in/id/eprint/2490 |
Actions (login required)
 |
View Item |
