Paper to Present:

Predicting Enhancer-Promoter Interaction from Genomic Sequence with Deep Neural Networks

Project:

Topic: Building an existing module using deep learning framework

• Update 1:
• Create a CNN module using MNIST dataset in Keras (using Tensorflow as backend)
• Explore DanQ: A hybrid convolutional and recurrent deep neural network for quantifying the function of DNA sequences
• Resolve the library dependency issues of Theano, Keras and seya to Run DanQ. Here Theano is used as backend of Keras
• Update 2:
• Exploring HYBRID SPEECH RECOGNITION WITH DEEP BIDIRECTIONAL LSTM
• Update 3:
• Spliting the 3.5G dataset into small part and run a portion of the full model on a small dataset
• Working on the libraray issue of forward, backward and bidirectional LSTM
• Update 4:
• Resolve all lib dependencies for forward, backward and bidirectional LSTM
• Run the train model for small portion of dataset which takes about 3 hour