SWaV: Unsupervised Learning of Visual Features by Contrasting Cluster Assignments (Mathilde Caron)

[@louislouis7388]

Thanks for sharing the inspirational talks with nice speakers. I have read their paper prior, after listening this conversation, I can understand more from their work.

[@sonOfLiberty100]

I would say you are the best machine learning podcast on youtube or even the world. Really good work

[@deepblender]

Love those conversations! Besides getting an overall technical understanding, hearing about the intuitions is invaluable! As she mentions a few times, deep learning is still mostly empirical. Ablation studies for all aspects are often not practical or possible, at least in my case. As such, having reasonable intuitions is very important.

[@MachineLearningStreetTalk]

Thanks for the comment!!

[@Fordance100]

Amazing paper and great conversation, I learned a lot from Mathilde Caron and other guests.

[@JuliusUnscripted]

Thanks for this really interesting conversation and great work! :) Impressive!

[@aritraroygosthipaty3662]

This is great.

[@maxgriffiths6968]

Excellent content

[@rajne.k9577]

9:54 💕

[@hejarshahabi114]

I am interested in swAV, but I do know how to implement it with my custom dataset(satellite imagery ) for image segmentation. Also, I would like to know if the trained swAV model will act as a backbone for my primary classification/segmentation model.

[@syedabdul8509]

why does terrace in Swiss have so many pebbles(stones) laid??

[@syed1297]

Love the conversations but I think author's part with slides would have been much better.

[@Fordance100]

There is a different one on youtube she did with pythorch lightning people with slides. I think this one actually provides more conceptual insight of the paper. I will read her paper and the article that reproduced the work. Personally, I think I need to invest at least 1 or 2 days to digest the paper and have good understanding of it.

[@crispr8742]

343