#044 - Data-efficient Image Transformers (Hugo Touvron) episode artwork

EPISODE · Feb 25, 2021 · 52 MIN

#044 - Data-efficient Image Transformers (Hugo Touvron)

from Machine Learning Street Talk (MLST)

Today we are going to talk about the *Data-efficient image Transformers paper or (DeiT) which Hugo is the primary author of. One of the recipes of success for vision models since the DL revolution began has been the availability of large training sets. CNNs have been optimized for almost a decade now, including through extensive architecture search which is prone to overfitting. Motivated by the success of transformers-based models in Natural Language Processing there has been increasing attention in applying these approaches to vision models. Hugo and his collaborators used a different training strategy and a new distillation token to get a massive increase in sample efficiency with image transformers.  00:00:00 Introduction 00:06:33 Data augmentation is all you need 00:09:53 Now the image patches are the convolutions though? 00:12:16 Where are those inductive biases hiding? 00:15:46 Distillation token 00:21:01 Why different resolutions on training 00:24:14 How data efficient can we get? 00:26:47 Out of domain generalisation 00:28:22 Why are transformers data efficient at all? Learning invariances 00:32:04 Is data augmentation cheating? 00:33:25 Distillation strategies - matching the intermediatae teacher representation as well as output 00:35:49 Do ML models learn the same thing for a problem? 00:39:01 How is it like at Facebook AI? 00:41:17 How long is the PhD programme? 00:42:03 Other interests outside of transformers? 00:43:18 Transformers for Vision and Language 00:47:40 Could we improve transformers models? (Hybrid models) 00:49:03 Biggest challenges in AI? 00:50:52 How far can we go with data driven approach?

Today we are going to talk about the *Data-efficient image Transformers paper or (DeiT) which Hugo is the primary author of. One of the recipes of success for vision models since the DL revolution began has been the availability of large training sets. CNNs have been optimized for almost a decade now, including through extensive architecture search which is prone to overfitting. Motivated by the success of transformers-based models in Natural Language Processing there has been increasing attention in applying these approaches to vision models. Hugo and his collaborators used a different training strategy and a new distillation token to get a massive increase in sample efficiency with image transformers.  00:00:00 Introduction 00:06:33 Data augmentation is all you need 00:09:53 Now the image patches are the convolutions though? 00:12:16 Where are those inductive biases hiding? 00:15:46 Distillation token 00:21:01 Why different resolutions on training 00:24:14 How data efficient can we get? 00:26:47 Out of domain generalisation 00:28:22 Why are transformers data efficient at all? Learning invariances 00:32:04 Is data augmentation cheating? 00:33:25 Distillation strategies - matching the intermediatae teacher representation as well as output 00:35:49 Do ML models learn the same thing for a problem? 00:39:01 How is it like at Facebook AI? 00:41:17 How long is the PhD programme? 00:42:03 Other interests outside of transformers? 00:43:18 Transformers for Vision and Language 00:47:40 Could we improve transformers models? (Hybrid models) 00:49:03 Biggest challenges in AI? 00:50:52 How far can we go with data driven approach?

NOW PLAYING

#044 - Data-efficient Image Transformers (Hugo Touvron)

0:00 52:22

No transcript for this episode yet

We transcribe on demand. Request one and we'll notify you when it's ready — usually under 10 minutes.

French Your Way Jessica: Native French teacher founder of French Your Way Boost your French listening skills and test your comprehension with this one of a kind series of podcasts. Get the chance to listen to a real conversation between native speakers talking at normal speed AND customise your learning experience through carefully designed sets of questions (2 levels of difficulty) available for download at www.frenchvoicespodcast.com. All interviews also come with the transcript. French teacher Jessica interviews native speakers of French from around the world who share a bit of their life and passion. Where else would you meet in one same place a French yoga teacher based in Melbourne, a soap manufacturer from Provence, or a couple cycling around the world? Kaizen Blueprint Aldo Chandra "Kaizen" is a Japanese term for continuous improvement. This podcast provides a blueprint to learn about health, wealth, relationships and everything else in between. Through our podcast, we strive to inspire, educate, and motivate our audience to cultivate a mindset of lifelong learning, productivity, and personal development. By sharing insights, strategies, and practical tips, we aim to guide listeners on their journey towards realizing their fullest potential, fostering success, and creating lasting positive change. One Man Went To Row PepperDawesMedia Follow the journey, from training to finish line, of a man from Derby, UK who is going from having only ever rowed on a machine to rowing 3000 miles solo across the Atlantic...just after his 70th birthday! Humanizing Change Tremendousness Join us each episode as we talk with innovators in their respective fields about their unique journeys and how they humanize change in their own work, right here, on Humanizing Change.

Frequently Asked Questions

How long is this episode of Machine Learning Street Talk (MLST)?

This episode is 52 minutes long.

When was this Machine Learning Street Talk (MLST) episode published?

This episode was published on February 25, 2021.

What is this episode about?

Today we are going to talk about the *Data-efficient image Transformers paper or (DeiT) which Hugo is the primary author of. One of the recipes of success for vision models since the DL revolution began has been the availability of large training...

Can I download this Machine Learning Street Talk (MLST) episode?

Yes, you can download this episode by clicking the download button on the episode player, or subscribe to the podcast in your preferred podcast app for automatic downloads.
URL copied to clipboard!