(3) Write down 2 applications of ANN for classification problems in Petroleum Engineering. 10:53 Answer 1 of 1 Done Part A Only one hidden layer is present in network Part: No of neurons - No of neurons in hidden layer No of neurons in output layer = 2+1=3 Part: No ac connections No ot connection from input 10 hidden layer + No of connection from hidden layer to output layer = 6+2=B Part: As there are two neurons in hidden layer, let M be the uuper one neuron in hidden layer neuron and h2 be the lower one neuron in hidden layer nouron Bias = 0.1 in a neurons and activation function is sigmoid Sigmoid = 10.5 0; itherwise Input at ht=34*0,62*0.1 +4-07 bias= 23.5 Output of htSigmoid (input) Sigmoid(23.5) Input at 12 = 34-02-20.3 + 402 bis 83 Output of h2 = Sigmoid(Input at h2) = Sigmoid(8.3) 1 Input at output neuron = 10.5 10.8 bias = 14 Fianloutput Sigmoid input toput neuro) Sigmoid 1.41 Network output=1 Parte As the final output is 0.2 which means we need to do the task of regression which means activation at each neuron will be linear only Output of ht=Input at ht: 34-0.6+20.12 40.7 23.4 Output of h2 = Input at 12 = 34'02. 20.3. 4-02 = 8.2 Fani output Input at output neuron = 23.405+ 8.2 0.8 = 18.26 Loss Finaloutput - Oroun truth = 18-28-02= 18.06 Let wights: W06,w203,w3=0.7,w= 02, w5 = 0.3, w6 = 02, W7=0.5, B = 0.8 We need to update w which is 0.3 1361 +2w2w3 h234w4.2w5w6 Output = w7Output of 11+ w&Output of hz We need to find gradien of w M. As GT is constant so, derivative of loss w.tw - derivative of output w.. w51 doutput/012 012/2 Therefoe, Joutput 2-0.8.2-16 Let learning rate = 0.1 By gradient descent W5w5-derivative of wloss -0.3 -0.116 18.06 = -25896 Note: If you have any doubt thon let me know by doing comment