Deep Parkinson Disease Diagnosis: Stacked Auto-encoder
In this work, we demonstrate the feasibility of deep learning based stacked auto-encoder for the Parkinson’s Disease (PD) diagnosis. Features are extracted by the employed deep network from the input source. We transfer features learned from the SAE during pre-training to the fine-tuning phase in which each sample or patient’s condition is labeled, which grants the network the time to learn and distinguish the healthy from the PD patients. The employed model is fine-tuned and tested on a small dataset in order to explore their generalization capabilities when trained using few data. Experimentally, the stacked auto-encoder showed a high accuracy and features extraction capability in diagnosing the Parkinson diseased patients where it achieved an accuracy of 89.5% which is considered as a promising result.
KeywordsDeep learning Stacked auto-encoder Parkinson’s disease
- 5.Gil, D., Johnson, M.: Diagnosing Parkinson by using artificial neural networks and support vector machines. Global J. Comput. Sci. Technol. 9, 63–71 (2009)Google Scholar
- 7.Helwan, A., Uzun, D., Abiyev, R., Bush, J.: One-year survival prediction of myocardial infarction. Int. J. Adv. Comput. Sci. Appl. 8, 173–178 (2017)Google Scholar
- 8.Ozsahin, D., Isa, N., Uzun, B., Ozsahin, I.: Effective analysis of image reconstruction algorithms in nuclear medicine using fuzzy PROMETHEE. In: 2018 Advances in Science and Engineering Technology International Conferences (ASET) (2018)Google Scholar
- 9.Khemphila, A., Boonjing, V.: Parkinsons disease classification using neural network and feature selection. World Acad. Sci. Eng. Technol. Int. J. Math. Comput. Sci. 6, 377–380 (2012)Google Scholar
- 10.Baldi, P.: Autoencoders, unsupervised learning, and deep architectures. In: Proceedings of the 2011 International Conference on Unsupervised and Transfer Learning Workshop, UTLW 2011, vol. 27, pp. 37–50 (2012)Google Scholar