Source: Deep Learning on Medium
Target: How to find the loss function of given data for the MP model
firstly we need to decide the threshold value, to get acquainted with the threshold value, you can refer https://medium.com/@ravichandrareddya9/how-mp-neuron-can-process-the-task-for-certain-data-dd0f84ae7038?source=your_stories_page—————————
you can evaluate loss by analysing the difference between like(y) and prediction (y_hat). But here the problem is by taking aggregation of individual loss we will get zero. So it’s not a good approach.
solution: taking a square to individual loss and aggregate.
you will get 1+1+1+1=4 as a loss. But here we can conceive another solution as |loss| (mod), then why we are approaching through square?
One answer for that is, for mod function differentiation we can’t define every value.