Скачать или смотреть 11. Maximum Likelihood (MLE) and Kernel Density Estimation (KDE)

This lecture series on maximum likelihood estimation (MLE) and kernel density estimation (KDE) explains how to estimate probability distributions from data. MLE finds the parameters of a known distribution that best fit the data by maximizing a likelihood function, solvable either through direct calculation or gradient descent. KDE, in contrast, constructs a distribution from data without assuming a specific form, using a sum of Gaussian distributions centered on the data points. The lecture uses examples and exercises to illustrate these methods, including calculating the probability of events and making predictions. The accompanying slides visually depict the methods and their application to various distributions. Both MLE and KDE are presented as tools for moving from data to a probability distribution, each with its own advantages and disadvantages.