Bruno Gavranović

The essence of the work that I am doing can be encapsulated under the umbrella term categorical cybernetics. Most of these papers assume knowledge of category theory, most notably that of lenses/optics. We propose a categorical framework for ‘cybernetics’: processes which interact bidirectionally with both an environment and a ‘controller’. Examples include open learners, in which the controller is an optimiser such as gradient descent, and open games, in which the controller is a composite of

3 mentions: @bgavran3@bgavran3@justinrwlynn
Date: 2021/06/14 15:19

Referring Tweets

@bgavran3 This is just some stuff from some of the recent papers I coauthored ( where this is all written down in more detail. The more I work on understanding learning through the lens of category theory, the more excited I get: I believe the potential is immense.
@bgavran3 @sam_power_825 Yeah, I think there's lots of categorical background that I did not mention in this tweet. This blog post is a short and down-to-earth introduction to these thoughts: and some of my recent papers too:
@justinrwlynn @tweagio Type Systems as Macros [System-Fω Powerful] Dependent Type Systems as Macros Neural Networks, Types, and Functional Programming Categorical Foundation of Gradient Based Learning

Related Entries

Read more Neural Networks through the Lens of Category Theory - YouTube
0 users, 4 mentions 2021/02/20 00:52
Read more GitHub - bgavran/autodiff: Rudimentary automatic differentiation framework
0 users, 1 mentions 2021/04/15 22:49
Read more 21w5251: Tangent Categories and their Applications (Online) | Banff International Research Station
0 users, 1 mentions 2021/06/17 10:49
Read more Video: Bruno Gavranovic, "(with Paul Wilson) Categorical Foundations of Gradient-Based Learning"
0 users, 1 mentions 2021/06/27 13:48
Read more Understanding Deep Learning
0 users, 1 mentions 2021/07/02 16:48



安定したサイト運営のためにGitHub sponsorを募集しています。