Effective classification of Focal class (FC) and Non-Focal Class (NFC) of Electroencephalography (EEG) is important to analyze patients’ brain activity for better diagnosis of epilepsy. For effective classification, a Deep Learning (DL) based classification of EEG signals for detecting FC and NFC is proposed in this research. The input EEG signals from the Bern Barcelona (BB) dataset are pre-processed using Independent Component Analysis (ICA) and Canonical Correlation Analysis (CCA) to eliminate noise and artifacts in the data. Three non-linear entropy-based extraction techniques are used, namely Approximate Entropy (ApEn), Sample Entropy (SampEn), and the Logarithmic-energy Entropy (LogEn). The extracted features are then fed as input to the ConVnet Bi-LSTM, which is a fusion of Convolutional Neural Networks (CNNs) and Bidirectional Long Short-Term Memory (Bi-LSTM). The CNN layers focus on finding the spatial relationship among signals, and Bi-LSTM considers the data in temporal directions, resulting in better classification of FC and NFC signals. Overall, experiment results show that the proposed approach achieved classification accuracy of 99.67%, precision of 97.95%, and F1-score of 97.17% when compared to results of existing methods, Recurrent Neural Network with Long Short-Term Memory (RNN-LSTM) and RNN with Gated Recurrent Unit (RNN-GRU).