If yes, can you suggest me some references to better understand the algorithm?
Suppose that I have 3 classes: dog, cat, horse.
I am not able to generalize what is explained in the lectures, how can I calculate entropy with more than 2 classes?
Generally, for each sample, you calculate one negative log probability of the true class the sample belongs to, then you add up all negative log probabilities.
Please google “cross-entropy” or with additional keywords such as “3 classes example”.