R/loss.R
crossentropy.Rd
binary cross entropy
crossentropy(y, p)
true label (0/1)
predicted probability
-(y * log(p) + (1 - y) * log (1 - p))