I was going through backpropagation in time for RNN, in the deep learning book of Joshua Bengio et.al. (deep learning book , section 10.2.1 ).

Given a network as:

enter image description here

the book tells that the backpropagation of the error gradient respect with the state h(t) is given by:

enter image description here

So, why is the matrix of the D(h_(t+1))/D(h_t) diagonal? How is its solution obtained?


I'll provide a sketch of the derivation. Omitting the bias term (since anyways we take derivatives later), the recursion looks like: $$\mathbf{h_{t+1}}=tanh(\mathbf{ U x_{t}+W h_{t}}) $$ where the $tanh$ is taken elementwise.

Now, since $\mathbf{h_{t}}$ and $\mathbf{h_{t+1}}$ are vectors, the derivative $\frac{\partial \mathbf{h_{t+1}}}{\partial \mathbf{h_{t}}}$ is a Jacobian.

  1. Notice that if $y=tanh(x), dy/dx=1-tanh^2(x)=1-y^2$.
  2. Let's see how a single element of the Jacobian looks like. Assume that the hidden layers are of dimension $n$ Now $h_{t+1, i}=tanh(\sum_{k=1} ^{n} w_{ik}h_{t,k}+ g(x))$. Here $g(x)$ stands for some function of x.

  3. Hence $\frac{\partial h_{t+1, i}}{\partial h_{t, j}}$ = $1-h_{t+1,i}^{2}w_{i,j}$. This corresponds to element in position $(i,j)$ of the Jacobian (from #1 above).

  4. This multiplier $1-h_{t+1,i}^{2}$ applies to each element of the $i^{th}$ row of the Jacobian.
  5. From basic linear algebra you can show that pre-multiplying a matrix $A$ by a diagonal matrix $diag(b_{i,i})$ is equivalent to scaling up row $i$ of matrix $A$ by $b_{i,i}$

Hence the term $diag(1-\mathbf{h_{t+1}^{2}})$ occurs as a premultiplier to the matrix $\mathbf{W}$

  • $\begingroup$ Could you please share how the derivation is performed? I can not understand form where come term W(i, j) ? if dy/dx=1-y^2 it needs to be just 1-h^2(t+1, i) I do not understand how the W(i,j) appears at the point 3. $\endgroup$ Feb 13 '18 at 9:08

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.