EEG Analysis to Decode Human Memory Responses in Face Recognition Task Using Deep LSTM Network

This chapter introduces a novel strategy to classify the human memory response involved in the face recognition task by utilizing the event‐related potential (ERP) signals. It introduces the composite effect of amplitude and phase angle of the signal in common spatial pattern (CSP) formulation and is solved using Lagrange's multiplier method, taking phase information of electroencephalography (EEG) into account. The chapter aims at designing a deep LSTM network with attention mechanism for classifying the human memory response captured by an EEG device during recognition of familiar and unfamiliar faces by utilizing ERPs. It presents a novel approach to utilize the composite benefits of deep LSTM and attention mechanism to realize the classification on a structural network. The chapter presents design issues of the proposed CSP for electrode selection and LSTM network for classification. It gives experimental details with detailed analysis of results.