Problem 58887. Neural Net: Best Index of Prediction array

Richard Zapor

18 solvers

1 likes

Solve

Solve Later

The final ouput of a neural net application is a Prediction matrix/vector,P, that gives the probability of the input being of a set of N classes represented by the columns.
For the XOR function the input matrix is [0 0;0 1;1 0;1 1] with an expected output vector of [0;1;1;0]. Ouput of Zero would be represented by a higher likelihood in column 1 than column 2 of the Prediction matrix.
Given a P matrix [m,n] where there are m cases and n possible outcomes return peak value for each case and the index of the most likely outcome.
[case_peak,case_idx]=Prediction(P)
if P was [0.1 0.5 0.9;0.2 0.8 0.4] (2 cases with 3 possible outcomes 1:3) return case_peak=[0.9;0.8] and case_idx=[3;2]

Solve