Updated Math, Added Component Flexibility

[iSiddharth20]

• Flexibility to Add different independent Datasets for AutoEncoder and LSTM, now these components don't necessarily have to be trained on the same images.
• Ability to define independent and different batch size for AutoEncoder Training and LSTM Training.
• SSIM will now act as Regularization Term along with MSE in SSIMLoss Function.
• Updated and Corrected Math for MaxEnt Regularization Term.
• For LSTM Training, Each Individual Sequence will now be split to create Training/Validation Datasets.