6.034 Artificial Intelligence - Recitations, fall 2004 online slides on learning
	Example If we use g(x) = x, g'(x) = 1, we let n = 1 ^ | | y = 5 t = 10 | +----------+ | | | NODE 1 | +----------+ ^ ^ ^ / | \ 5/ 2| \5 / | \ +-+ +-+ +-+ |2| |3| |4| (Output from these nodes is +-+ > +-+ < +-+ 0, 0, 1) ^ / ^ \ ^ | / | \ | |1 /1 |1 0\ | 1 | / | \ | | / | \ | +-+ +-+ +-+ |5| |6| |7| +-+ +-+ +-+ ^ ^ ^ | | | 0 0 1 delta_1 = (10 - 5) * 1 = 5 w_21 = 5 + 1*0*5 = 5, w_31 = 2 + 1*0*5 = 2, w_41 = 5 + 1*1*5 = 5 + 5 = 10 delta_2 = 1 * w_21 * delta_1 = 1 * 5 * 5 = 25 delta_3 = 1 * w_31 * delta_1 = 1 * 2 * 5 = 10 delta_4 = 1 * w_41 * delta_1 = 1 * 5 * 5 = 25 Therefore w_52 = 1 + 1*0*25 = 1, w_53 = 1 + 1*0*25 = 1 w_63 = 1 + 1*0*10 = 1 w_73 = 0 + 1*1*10 = 10, w_74 = 1 + 1*1*25 = 26 Now for same values from input nodes (0, 0, and 1), output will be 10*2 + 26*5 = 150, which is a move in the right direction but too big a move.