Set up your training data in an Excel sheet spanning columns A, B, and C: Target Output 2. Initializing Weights and Biases

Excel has a built-in optimization engine called that can automate this loop instantly.

We will build formulas for the first training row (X1=0, X2=0, Target=0) in columns F through K.

After Solver finishes, the "Total Error" should be very low (e.g.,

We pass the output gradient backward through the weights to find the error responsibility of each hidden node: =$U2*$L$2*N2*(1-N2) Cell W2 ( δh2delta sub h 2 end-sub ): =$U2*$M$2*O2*(1-O2) Cell X2 ( δh3delta sub h 3 end-sub ): =$U2*$N$2*P2*(1-P2) Step 3: Weight Gradient Calculations