Variation in Neural Network Weights if training data is shuffled?

Question

We have two copies of the same Neural Net (NN). Lets call them NN1 and NN2.

NN1 is trained on a Training set (TSo) that yields weight set W1

If we train NN2 on TSo's shuffled set (TSs), will the weights yielded, W2 be same as W1?

Data to be trained is same, just the ordering of the training data is different.

NN1 = NN2 (Initially)
NN1 ---> TSo ---> W1
NN2 ---> TSs ---> W2
Is W1 same as W2? Why or Why not? 

Answer 1

Weights will not be the same.

Weight updates are dependent on the current batch of images. If you're using shuffled dataset, the ordering of images in the batches are different. Thus gradients will be different & weights updates will be different.