Power K-Means
Data Skeptic
00:00
The Problem of Local Minimization in Machine Learning
K means is a prett famous algorithm. A lot of people, their first experience is day one of their first machine learning class. The simple te state objective is actually quite difficult to solve in sort of a head on way. So lloyd's algorithm will converge quickly, often, as you said, even for large data sets. But if you re started that algorithm from multiple random initializations, you can get different answers.
Transcript
Play full episode
