If you watch the next video in the playlist, you see there appears to be little content missing from this video. The next video begins with a review and then a quick addition of the demonstration of the usage of RBFs still results in a linear squares problem (pages 11 and 12 of the corresponding class notes).
Does anyone have an intuitive idea about why Shannon entropy takes the order of a probability as the weight of the probability itself? Is that related to softmax?
Entropy is expected value of surprise. SO less likely events will have high surprise . That is why inverse of a proability of an event is taken as a measure of surprise. And for practical reasons we take log of the inverse of the probability. Now we calculate the expected value of the surprises to get the entropy. Expected value is nothing but a weighted average wherein weights are probabilities themselves.
this professor is amazing
If you watch the next video in the playlist, you see there appears to be little content missing from this video. The next video begins with a review and then a quick addition of the demonstration of the usage of RBFs still results in a linear squares problem (pages 11 and 12 of the corresponding class notes).
Thanks for uploading it Professor de Freitas !
Hy Nando, where is the continuation of this video? thank you!
I wish if i discovered your courses a bit earlier
The video cuts off unexpectedly at the end.
Great lecture!
Nando if you have given a lecture on Support Vector Machine...Can you kindly share it too.
Does anyone have an intuitive idea about why Shannon entropy takes the order of a probability as the weight of the probability itself? Is that related to softmax?
Entropy is expected value of surprise. SO less likely events will have high surprise . That is why inverse of a proability of an event is taken as a measure of surprise. And for practical reasons we take log of the inverse of the probability. Now we calculate the expected value of the surprises to get the entropy. Expected value is nothing but a weighted average wherein weights are probabilities themselves.
@@kurienabraham8739 Thank you.
@@BruinChang Never mind. I learnt it from here - th-cam.com/video/YtebGVx-Fxw/w-d-xo.html
The curve in 50:20. Is it always a single line. I feel like a line made by two (2D) spheres will be two curves. E.g. looks like you draw an eye.
the lecture is missing some content at the end after "how do we compute ...."