A collaborative list of interactive Machine Learning, Deep Learning and Statistics websites. Started by Piotr Migdał, but anyone is encouraged to contribute! It is a simple no-build Vue.js website:
Feel invited to Pull Request other interactive visualizations (check websites.yaml)! :)
...aaand if you want to create such visualizations by yourself, see In Browser AI.
Still I am thinking what is the best criterion.
For sure things that are front-end (i.e. JavaScript within browser). For things using backend (when you can see solution, but it uses some PyTorch/TF/etc code on a server) I am still debating, but I lean on being more inclusive. In this context:
- make sure it has some didactic value (otherwise ALL services using ML would qualify)
- add
backend-dependentinuses
Strong preference for open-source solutions (so people can reuse it and learn from code), though it is not a requirement. Though, mention repo and open source license only when it is directly relevant (vs additional materials such as exercises for a book, or Python algorithm).
Read Explorable Explanations by Bret Victor.
Inspirations for collecting and displaying content:
- Science-based games - a collaborative list - a list I started (maybe I will turn it int something interactive as well)
- Kaggle Past Solutions - a searchable compilation of Kaggle past solutions
- D3 Discovery - finding D3 plugins with ease
Main layout and styling developed by Jakub Fogel
(You are invited to constribute)
- Descriptions of sites
- Write-up in a different way
- Some sorting (alphabetical?)
- Share button
- Code refactoring :)