I. Introduction
The neurotransmitters in the brain are composed of gamma-aminobutyric acid (GABA) and glutamate[1]. These neurotransmitters create an imbalance in the brain resulting in a huge flow of sodium and calcium ions. These ions regulate the aspartate excitatory and glutamate neurotransmitters. Rapid changes occur in the potassium and chlorine levels responsible for interceding GABA levels, thereby reducing inhibitions [2]. Increased excitatory and depleted inhibitory neurotransmitters lead to ahigh degree of excitability and hyper-synchronization of neurons. These activities spread throughout the brain, resulting in a seizure. Seizures are divided into Partial and Generalized seizures, which is further sub-categorized based on the location of the seizure, visual symptoms, the awareness level of the patient, etc. The seizures have four states: i) pre-ictal state precedes the actual seizure; ii) ictal state is the actual seizure, iii) post-ictal state is just after the seizure, iv) inter-ictal is the normal state or between seizures. For the efficient diagnosis and accuracy of detection of epileptic seizures, the omni present background activities need to be differentiated from the epileptic patterns. Different machine learning techniques have been adopted to classify the EEG signals accurately. Feature extraction and its classification are critical for accurately detecting epileptic seizures. In this paper, we discuss the outcome of adopting machine learning classifiers like K-NN and SVM for achieving high efficacy and accuracy in epileptic seizure detection. The research has been undertaken on the database of the University of Bonn available on public domain.