Hi All,
This week the tea talk will take the form of a lab-wide discussion of our future ... code-wise. Hope to see you there.
When: Thursday Sept. 2nd, 14h30 - 15h30 (Tomorrow) Where: LISA Lab.
Title/Abstract:
Brainstorming on the future of shared machine learning code on top of Theano
The objective is to hear every one that has some opinion on this subject, mostly people who have designed or thought about learning frameworks on top of Theano, or people who know about other machine learning libraries from which we could learn something. There are those that lean towards a well-organized *library* defining functional types (such as types of datasets, of learners, of optimizers, of performance statistics, etc.) and those that believe that a more organic sharing of pieces of code would work better. But if we are going to share code and make it easy to REUSE pieces already written by others, we have to start agreeing on conventions and protocols, at least, for the various kinds of pieces one may find in this repository, which would become the successor of Pylearn.
Cheers, Aaron
Afficher les réponses par date