Optimization of Machine Learning Methods for Robust Biological Image Analysis

Generate ground truth image sets to train the machine learning model
Advance the prediction pipeline to provide accurate results with multiple types of input images

Developing MATLAB script to generate ~100 ground truth image masks for a machine learning training set
Learning code base, machine learning libraries (Keras and Tensorflow), model parameters, loss functions and optimizers to better understand and implement changes in the model
Updating code to reflect most recent Keras-Tensorflow integration
Utilizing remote shared computing cluster to make training 12 times faster and expand model capabilities
Utilizing TensorBoard library to visualize loss and accuracy per epoch
Successfully generalizing model to predict more accurate results in a mixed image dataset