Question

Gradient descent weight update rule for a tanh unit. (2 pts) Assume throughout this exercise that we are using gradient descent to minimize the error as defined in formula (4.2) on p.89 in the textbook: td -od Recall that the corresponding weight update rule for a sigmoid unit like the one in Figure 4.6 on p.96 in the textbook is: td - od) od (1-od) i,d ded Let us replace the sigmoid function σ in Figure 4.6 by the function "tanh". In other words, the output of the unit is now: o = tanh(net) = tanh(uoro + wJZ1 + tv2x2+ + u'nxn) Derive the new weight update rule. Show your work, and indicate clearly in your answer what the weight update rule for the tanh unit looks like. Hint: tank(z) = 1-(tanh(z))2. 지 wo x2 net .ari 'i ン..ornet) ...._ 1+e Ww PI FIGURE 4.6 The sigmoid threshold unit.

Answer 1

IF YOU HAVE ANY DOUBTS COMMENT BELOW I WILL BE TTHERE TO HELP YOU..ALL THE BEST..

AS FOR GIVEN DATA.

The cncr atoudput nod and dden node sea t? → The "ta(get output eran it ourput ot unt > haina vue ret Cs Scanned CamScanner ret

hidken uoits *cünins neu ded Bnetk do Z)と A0 Ke donseam Scanned with CamScanner

I HOPE YOU UNDERSTAND..

PLS RATE THUMBS UP..ITS HELPS ME ALOT..

THANK YOU...!!