Java Deep Learning Projects

Cancer type prediction using recurrent type networks

We solved an interesting project, where we successfully classified cancer patients based on cancer types. For this, we used the LSTM network. We used a very high-dimensional gene expression dataset. We converted the dataset into sequence format and trained the LSTM net for each sample per time step.

This project also shows the robustness of deep architecture such as LSTM, demonstrating that even without applying dimensionality reduction, the model can handle a very high-dimensional dataset.

One of the potential limitations of this approach was that we considered only a gene expression dataset, so it cannot be deployed for a real-life prognosis and diagnosis, whereas other datasets such ...