The estimator is obtained by solving that is by finding the parameter that maximizes the log-likelihood of the observed sample. As a family does not have to specify how to calculate the log-likelihood this is based on the familys function to compute the AIC. Li f k Li pk pk f k 1 pkpk 1 pk pk 1 L i f k L i p k p k f k 1 p k p k 1 p k p k 1 And thus we have differentatied the negative log likelihood with respect to the softmax layer.
Write the natural log.
This is simply the product of the PDF for. Li f k Li pk pk f k 1 pkpk 1 pk pk 1 L i f k L i p k p k f k 1 p k p k 1 p k p k 1 And thus we have differentatied the negative log likelihood with respect to the softmax layer. I think which the fitdist function includes likelihood functioneg. 3 -- Calculate the log-likelihood.