Efficient classification of Parkinson's disease using extreme learning machine and hybrid particle swarm optimization

One of the most well-known problems in machine learning framework is classification of Parkinson's Disease (PD) to patient people and healthy people. Due to the importance of that problem, utilization of a novel learning method is necessary. For this purpose, this paper proposes the utilization of Extreme Learning Machine (ELM) as a type of feed-forward neural network with a single hidden layer to classify the PD patients. However ELM is known as the one of the fast and accurate learning methods, selection of relevant feature elements of PD dataset can be effective on improving the classification performance of ELM. To this end, this paper proposes Hybrid Particle Swarm Optimization (PSO) as the second innovation to efficiently select the relevant feature elements. The main advantage of Hybrid PSO is locally improving of particles in order to jump over the local optimum solution and quickly converging to the global optimal solution. Evaluation of the proposed method on PD dataset proves the superiority of the propose method on the problem of PD classification, in comparison to the other learning methods.