in neural networks, what happens if we interchange z[l]=w[l]a[l−1]+b[l] w and a terms 
z[l]=w[l]a[l−1]+b[l]

In this if I interchange w and a terms i.e.

z[l]=a[l−1]w[l]+b[l]

I know matrix multiplication constraints that is no. of columns in first matrix should be equal to number of rows in second matrix.
If we ignore those constraints, that is these constraints are satisfied

z[l]=a[l−1]w[l]+b[l]   is valid,

then how will this affect my neural network?
Will this work?
what change can I expect?
 A: Assuming that you are refering to a fully connected neural network, the key issues to realize here are:

*

*$w^l\,\,\,\,\,\rightarrow$ is the matrix of weights that connect the neurons of the layer $l-1$ with the layer $l$. The most spread notation is to define it as a $\text{J}\times \text{K}$ matrix, where $\text{J}$ is the number of neurons of the layer $l$ and $\text{K}$ Is the number of neurons of the layer $l-1$.

*$a^{l-1}\rightarrow$ due to the previous notation of $w^l$, the vector of activations of the layer $l-1$ Is a $\text{K-dimensional}$ vector.

*$b^l\,\,\,\,\,\,\rightarrow$ Also, due to the previous notation, the vector of biases of the layer $l$ is a $\text{J-dimensional}$ vector.

*$z^l\,\,\,\,\,\,\rightarrow$ Has to be a $\text{J-dimensional}$ vector.

With this said, let's have a look at what you are proposing:
$$ z^l = a^{l-1}w^l + b^l$$
The only situation that I can come up with where this expression can work is if we are calculating each neuron's $z^l_j$ individually, and we also would have to change the usual notation $\rightarrow$ the former $a^{l-1}$ would have to be tranposed, and we would have to work with the vector of weights of neuron $j$ rather than with the weight matrix of layer $l$. Finally $b^l_j$ would have to be of course a scalar as it represents the bias of only the neuron $j$.
So to sum up, changing the usual notation is possible but it may become less efficient. As we have saw above, with the proposed equation, we can no longer work with matrices $\rightarrow$ the computation become more slower.
