● Researched on paper about ELM (Extreme Learning Machine). ● Researched how to inject ELM in different models architecture such as CNN or Transformers (Encoder and Decoder both) to see the speed improvement.

● Developed a model that determines the binding parameter of epitope-paratope pair connection using various neural networks structures such as CNN, LSTM, RNN, GRU. ● Used BERT transformer to encode the data of epitope and paratope acid sequences and hereby the model accuracy increased by 15%. ● Suggested to use SMOTE technique to make balance in the target data and therefore F-1 score of our model increased by 6%.