(Day 124) MLx Fundamentals Day 1 - Intro to ML, Naive Bayes, Factorization methods

Ivan Ivanov · May 4, 2024

Hello :) Today is Day 124!

A quick summary of today:

  • Introduction to Machine learning by Volodymyr Kuleshov from Cornell Univeristy
  • Naive Bayes practical session by Richard Willis from King’s College London
  • Factorization methods by Cho-Jui Hsieh from UCLA

image

Introduction to Machine learning by Volodymyr Kuleshov from Cornell Univeristy

It was really an amazing introduction. Personally it did not cover new stuff for me, but as an intro I believe it was top. What is supervised ML, OLS, Covered Non-Linear Least Squares, Overfitting, Regularization

Naive Bayes practical session by Richard Willis from King’s College London

The interesting bit was implementing a Naive Bayes classification model from scratch.

I feel like this is one of the simplest explanations of Bayes theorem that I have seen/read so far (of prior, likelohood and posterior). Richard Willis is a Phd

image image image image image

Implementation

Computing the prior

image

Computing the likelihoods

image

Computing log posterior

image

Complete NaiveBayesClassifier

image

Factorization methods by Cho-Jui Hsieh from UCLA

I went over the recording and took notes.

Covered topics:

Matrix factorization approach, Altering Least Squares, SGD, extreme multi-label classification, two-tower models, low-rankness for efficient DL, param efficient fine-tuning

image image image image image image image image

I am writing this blog in the break of Day 2. I will share what Day 2 of MLx Fundamentals is about tomorrow.

That is all for today!

See you tomorrow :)