arg min substack

Home
Lecture Blogs
Archive
About
New
Top
Discussion
The Deep Optimization Cookbook
Get those models to converge with one weird trick.
 • 
Ben Recht
7
No one knows how the brain works.
Is that why we believed it was hard to train neural networks?
 • 
Ben Recht
23
It's not a trick...
It's not an illusion either.
 • 
Ben Recht
14
The maddening nonlinear multiverse
The paths to nonlinear prediction all lead to the same place
 • 
Ben Recht
9
Hidden Foundations
We hide a lot of theory in our representation of data.
 • 
Ben Recht
5
Features of the foundations of machine learning
Which measurements let us predict?
 • 
Ben Recht
1
I don't know what I've been trying to prove.
Yeah, yeah. Why I feel like I do about generalization theory.
 • 
Ben Recht
29
Three paths to generalization
Uniform Convergence, Sequential Prediction, and Stability
 • 
Ben Recht
Generally speaking when god plays dice
An attempt at re-rethinking generalization theory
 • 
Ben Recht
1
Regretfully Yours
This is a live blog of Lecture 5 of my graduate machine learning class “Patterns, Predictions, and Actions.” A Table of Contents for this series is…
 • 
Ben Recht
4
Rigor vs. GitHub Descent
Is there a principled way to optimize in machine learning?
 • 
Ben Recht
22
7 Minute Opts
Optimization for Machine Learning In 1000 Words or Less.
 • 
Ben Recht
© 2023 Ben Recht
Privacy ∙ Terms ∙ Collection notice
Start WritingGet the app
Substack is the home for great writing