A generic Neural Network based Deep learning model has been evolved as a part of the Dissertation. The fashion data set has been chosen for the analysis in the Dissertation. The fashion data set has monochrome bitmap images of ten different fashion wears, due to this the fashion data set has Ten Classifications and the model needs to classify it correctly.

First a basic Neural Network without any Hidden Layers has been chosen for classification model. Later models with the Two Hidden Layers with Sigmoid/RELU and Softmax Activators, Adam Optimizer and Random Dropout have been analyzed. The test data accuracy and the cross Entropy has been chosen as the parameter for the comparison. It has been found that the Two Hidden Layers with the RELU and Softmax Activators along with the random Dropout has the best Performance. The Convolutional Neural Network model has been chosen for the analysis and found that it is better than the Two Hidden Layers with the RELU and Softmax Activators along with the random Dropout.