This was exactly the baby step I needed to get me on my way with entropy. Far too many people try to explain it by going straight to the equation. There's no intuition in that. Brilliant explanation. I finally understand it.
@@SerranoAcademy Repetition (redundancy) is dual to variation -- music. Certainty is dual to uncertainty -- the Heisenberg certainty/uncertainty principle. Syntropy (prediction) is dual to increasing entropy -- the 4th law of thermodynamics. Randomness (entropy) is dual to order (predictability) -- "Always two there are" -- Yoda.
how does one make something so complicated into something so intuitive that others can finally see the picture. your explanation itself is an amazing feat.
Excellent explanation, very clear and concise! I have always pondered the significance of the log in cross-entropy loss function. The explanation (particularly: "products are small and volatile, sums are good") completely clears this up.
I have been scared of delving into entropy in detail for so long because the first time I studied it, it wasn’t a good experience. All I want to say is THANK YOU!!!!!! I should have been supplementing the udacity ND lesson videos with these since the beginning.
I'm studying Decision Tree (Machine Learning Algorithm) and it uses Entropy to efficiently build the tree. I finally understand the details. Thank you!!
Good video! Minor correction of calculations: at 5:50, the probability of getting the same configuration is 0.25. This is because there are only 4 possible configurations of the balls (there is only one blue ball, and only four slots, so only 4 places the blue ball can be). This can also be calculated by selecting red balls first multiplying 0.75 * 0.66667 * 0.5 = 0.25. Similarly, at 6:58, the probability is 1/6 because there are 6 possible configurations. We can calculate the probability by multiplying (2/4) * (1/3) = (2/12) = (1/6) ~= 0.166667.
Thanks for the relationship between knowledge and entropy, that was very helpful. Your explanation of statistics is also good! Though, I am only half way through the video at this point, I will finish it! Thanks
Wow, thank you, man. I needed that information! There are many ways to teach the same stuff! That number of question stuff is great! It's good to have more than one way to measure something!
I find sum(p*log(p^-1)) more intuitive. Inverse p (i.e. 1/P) is the ratio of total samples to this sample. If you ask perfect questions you'll ask log(1/p) questions. Entropy is then the sum of these values, each multiplied by the probability of each, which is how much it contributes to the total entropy.
hi Luis, nice to meet you, I am reading the book of Deep learning of Ian Godfellow, and I needed to view your video for understand the chapter, 3.13 information theory. thanks very much.
Luis, Thank you so much for this brilliant elucidation of information theory & entropy. Merely as an avocation, I have been toying around with a pet evolutionary theory about belief systems and societies. In order to test it - if that is even possible - I felt I needed to develop some sort of computer program as a model. Since I have very little programming experience and only mediocre math skills, I have been teaching myself both (with a lot of help from the web). It was purely by accident that I stumbled upon Claude Shannon and information theory, and I immediately became fascinated with the topic, and have a hunch that it may somehow be relevant to my own research. Regardless, I am now interested in it for its own sake. I had a an ephemeral understanding of how all the facets (probability, logs, choices, etc.) were all related mathematically, but it wasn't until after watching your video that I believe I fully grok the concept. At one point early on, I found myself shouting, "if he brings up yes/no questions, I know I understand this!" And then you did. It was such a wonderful moment for someone who finds math so challenging, and it is greatly appreciated! I shall check out your other videos later. You're a very good teacher!
In the last minute of the video, he explains that using Log base 2 corresponds to the level of a decision tree, which is the number of questions you'd have to ask to determine a value.
Syntropy is dual to increasing entropy -- The 4th law of thermodynamics! Thesis is dual to anti-thesis -- The time independent Hegelian dialectic. Schrodinger's cat: Alive (thesis, being) is dual to not alive (anti-thesis, non being) -- Hegel's cat. Syntropy is the process of optimizing your predictions to track targets or teleological physics. Teleological physics (syntropy) is dual to non teleological physics (entropy, information).
Hi. Thanks a million times for simplifying a very complicated topic. Kindly find time n post a simplified tutorial on mcmc.... I am overwhelmed by your unique communication skills. Markov chain Monte Carlo. God bless you.
Fácil de compreender por causa da excelente explanação. Parabéns pelo vídeo e muito obrigado por compartilhar. / Easy to think because of the excellent explanation. Congratulations on the video and thank you very much for sharing.
This is the best explanation I have come across for a long time, Can you please answer how can we use entropy to find the uncertainty of a naive Bayesian classifier with let's say 4 feature variables and a binomial class variable?