Epilepsy is the most common neurological disease comprises a heterogeneous group of disorders, characterized by recurrent and unprovoked seizures due to the huge electrical discharges of large synchronized neurons. A major concern of patients with epilepsy, especially thosewith drug-resistant epilepsy, is the random and unexpected occurrence of epileptic seizures. The anticipatory anxiety, feeling of helplessness and restrictions of daily activities impair the quality of life of an individual with poorly controlled epileptic seizures. The detection of electrical onset of seizures have a great impact on the management of epilepsy. This thesis work analyzed the impact of normalization schemes, sliding window length selection, features selection and performance of machine learning algorithms for the design of patient independent system to detect the electrical onset of temporal lobe epileptic seizure. The proposed system does not require prior knowledge on channel selection, and data selection for training the classifier. This improves the comfort level of end users (Neurologists and Neurotechnologists).The automated seizure detection system can also help in delivering the stimulation in responsive neurostimulator system (RNS/closed loop interventions).