What is Zero-Shot Learning?

Want to play with the technology yourself? Explore our interactive demo → ibm.biz/BdKkPk
Learn more about the technology → ibm.biz/BdKky2
Humans can on average categorize 30,000 unique object categories, but how can AI systems distinguish between different things? Zero-shot learning can be done through various methods like attribute based and embedding based learning. In this video Martin Keen explains how technologists are giving AI systems the categorization power that we take for granted.
AI news moves fast. Sign up for a monthly newsletter for AI updates from IBM → ibm.biz/BdKkyq

[@aott6799]

Mr. Keen lives up to his name. Another great presentation 👍

[@raymobula]

A video about GANs would be great.

[@MartinKeen]

A GAN video was one of the first I ever recorded for this channel... such a fun algorithm: ruclips.net/video/TpMIssRdhco/видео.htmlsi=9tVig_Xhq2UufPbE

[@geniusxbyofejiroagbaduta8665]

Thanks i now have an understanding of this subject

[@naridabanucio2828]

Great teaching

[@hammadaslam9201]

Wonderful explanation of ZERO short learning,
thanks for sharing a quality content :)

[@The_Artist_Gt_plus]

I want to get invole in tech field!

[@bamh1re318]

Me too! But IBM educational videos seems to release enough information on their capabilities to set us up to employ their WastonX. There is no double it will become great agents for many business

[@SHANEALAM-p5x]

IBM 🎉🎉🎉🎉🎉🎉🎉🎉🎉🎉🎉GREAT

[@thesimplicitylifestyle]

GAN + RAG + Embodiment = AGI
😎🤖

[@FREELEARNING]

Thanks for the nice tutorial. while I think that the term "Learning" is kind of misleading. in reality the model's parameters are not changed at all. witch means that their is no leaning.
The only thing that happens is that we are giving the model some labeled examples to the model as part of our prompt to kind of guiding it how to respond to an unlabeled example.
it's like when I give you this sequence "1 2 3 4 5", and I say complete the sequence "7 8 9" by following the same logic. So you will say "7 8 9 10".

[@JustAThought01]

Thank you very much for this video. It is a great summary of the current state of neural network development.

[@HDennis]

He must be writing everything mirrored for us to see it correctly, right?

[@ambareesha04]

That would be incredibly hard to do so quickly and without errors. I’m wondering if the video itself is mirrored.

[@waynet285]

Hi Dennis. I think he's likely right-handed and writing normally, and they horizontally flipped the video in post-production so it's not back to front for viewers :) This video (though somewhat old now in internet terms) explains it well: ruclips.net/video/eVOPDQ5KYso/видео.html