Скачать с ютуб From God Equation to Linear Regression in Machine Learning (& Physics) в хорошем качестве

From God Equation to Linear Regression in Machine Learning (& Physics)

In this video, I will start with the God equation (a general form of probability) and derive the linear regression equation in machine learning. This equation also co-exists in physics but under a different name. I show that the maximum entropy principle, from the second law of thermodynamics for physical systems and from information theory for non-physical systems, plays a crucial role for our purpose. Since a system tends to evolve toward a maximum disorder (entropy) state, it is a steady state in most situations. The steadiness of the system allows us to keep only the first few terms of Taylor's expansion of the probability function. That is the reason the probability function takes a universal form in any system that is in its steady state. That in turn leads to the linear regression equation being a valid predictor of most systems that are widely different in nature. I hope you enjoy this video. Please let me know what you think in the comments. Chapters: 00:00 - Introduction 00:29 - Who I am 00:49 - Summary of Previous Video 01:19 - True Probability is Unknown 02:15 - Don't Need Exact Probability 03:24 - Concepts: First Example 05:04 - Concepts: Second Example 08:25 - Concepts: Third Example 09:13 - One Dimensional Distribution 09:46 - Higher Dimensional Distribution 10:15 - Machine Learning Conventions 11:11 - Rearranging Into One Dimensional Distribution 12:50 - Introducing a Genius