artifitial nueral netwrok problem

The input vector to neural network will have shape p=shapetextureweight An orange is round, rough and less than one pound so po =111 - An apple is round, smooth and less than one pound so pa=111 - The neural network should expect input po or pa as input Example Randomly initialize W and b - W=[0.510.5],b=0.5 - Calculate output for first input p1 - a= hardlim (Wp1+b)= hardiim [0.510.5]111+0.5=hardlim(0.5+1+0.5+0.5)=hardlim(2.5)=1 - Calculate error e=ta=01=1 - Update weights and bias -Wnew=Wold+epT=[0.510.5](1)[111]=[0.500.5]-bnew=bold+e=0.51=0.5 - This completes first iteration Calculate output for second input p2 - a= hardlim (Wp2+b)= - Calculate error e=ta=10=1 - Update weights and bias - Wnow=Wold+e=[0.500.5]+(1)[111]=[0.510.5] - bnew=boll+e=0.5+1=0.5 - This completes second iteration - Calculate output for first input p1 - a= hardilim(Wp1+b)= hardiim [0.510.5]111+0.5= hardlim (0.51+0.5+0.5)= hardlim (0.5)=1 - Calculate error e=ta=01m=1 - Update weights and bias - Wnew=Wila+ep1=[0.510.5](1)[111]=[0.5220.5] - bnew=boud+e=0.5+(1)=0.5 - Update weights and bias - Wnew=Wold+eg=[0.510.5](1)[111]=[0.520.5] - bnew=bold+e=0.5+(1)=0.5 - This completes third iteration - Calculate output for second input p2 - Calculate error e=ta=11=0 - Calculate output for first input p1 - a= hardlim (Wp1+b)= hardlim [0.520.5]1110.5= hardlim (0.520.50.5)= hardlim (3.5)=0 - Calculate error e=ta=00=0 - The algorithm has converged to a solution 9) Repeat the same calculation as in the example above by changing Randomly initialize W and b to W=[0.210.2],b=0.2 10) Repeat the same calculation as in the example above by changing Randomly initialize W and b to W=[0.810.8],b=0.8 11) What do you conclude