Neural Network Team

This is the home repo for the Neural Network team.

Team Members

Adam Stafford, Tianhao Zang

Objective

To explore the implementation of nucleotide data into a neural network.

To do

Priority Items

• Make subsets of sequences to do diagnostics on
• Diagnostics: Build nueral network and run
  • try a few different architectures, no more than three layers
  • assess time on different number of subsets, quantify and visualize
  • assess memory on different number of subsets, quantify and visualize

Other

• Explore other options for data structuring of nucleotide data beyond one hot encoding
  • read section 2.2 in Deep Learning in Python
  • Make notes on what you are thinking and upload to notes.md
• Identify network Architecture, which layers?
  • What sort of layers and in which confirmation are we going to use?
• Get data into one hot encoding format
  • classification 1: binary (presence or absence of GAL4 signal)
  • classification 2: classification of stage of signal

The data

The data is located on Google Drive. Please use data/3.species_24_only_07March2018

Key points to keep in mind

• Structual organization of nucleotide input must be upheld.

Platforms

Main Language: Python Implementation: TensorFlow, Keras

Data Management

• See data_management.md