Parameters dimension in neural network

Course Q&A Machine Learning Specialization Advanced Learning Algorithms
1

I have also taken the older machine learning course (stanford).
In that for layer j the dimension of weight matrix is s_(j+1) *s_j+1.
In the weight matrix first column represent the bias terms of each neuron in the layer j.

But in this course the weight matrix is of order s_j*s_(j+1) and bias of layer j is in different vector.

s_j is number of neuron in layer j.

Can’t we follow the older course representation of weight matrix?

2

In theory you can use whatever notation you like that matches the data set you’re given.

But in the programming assignments in this course, you need to use the conventions given here.

Note that Andrew switches notations for some of the assignments. This helps show that there is no single universal agreement on the orientation of the data or the weights.

3

@TMosh, Thank you very much for the explanation.