Explore advanced concepts in data science through a comprehensive lecture covering Singular Value Decomposition (SVD) applications, Gaussian mixtures, and stochastic modeling alternatives. Delve into numerical algorithms, randomized approaches, and low-rank approximations with additive error. Gain insights on data handling techniques and pass-efficient models. Conclude with an introduction to Markov Chains, examining conductance and rapid mixing in symmetric Markov Chains. Learn from Microsoft Research India's Ravi Kannan as he presents key topics from the Foundations of Data Science Boot Camp.
Overview
Syllabus
Intro
Application of SVD to Gaussian Mixtures
SVD subspace = space of means
Life without a Stochastic Model: An Example Theorem Hypothesis
Lile without a Stochastic Model An Example
Doing without a stochastic Model
Numerical Algorithms
Why Randomized Algorithms?
Simple Setting
Problems
A little Notation
Low Rank Approximation with Additive Error
Data Handling, Pass efficient Model
Length squared sample of rows and col's suffice
Different Topic: Markov Chains A Markov Chain (MC) is a directed graph with positive edge
Conductance, Rapid Mixing of Symmetric MC's
Brief Idea of Proof
Taught by
Simons Institute
