Superpowers for next gen web apps: Machine learning

Discover how to achieve superpowers by embracing machine learning in JavaScript using TensorFlow.js in the browser. Learn what machine learning is, get a high-level overview of how it works, get inspired through a whole bunch of creative prototypes (from invisibility to teleportation) that push the boundaries of what is possible in the modern web browser, and then take your own first steps with machine learning in minutes. By the end of the session, everyone (no matter what your background) will understand how to recognize an object of their choice, which could then be used in any creative way you can imagine on your own website. No background in machine learning is required. Take your first steps with TensorFlow.js!
Resources:
Learn more about TFJS → goo.gle/2UXmFzX
Get Inspired with top community projects → goo.gle/39kf2vE
#MadeWithTFJS latest submissions → goo.gle/377lUKe
Speaker: Jason Mayes
Watch all Chrome Developer Summit sessions here → goo.gle/cds20-sessions
Subscribe to Google Chrome Developers here → goo.gle/ChromeDevs
#chromedevsummit #chrome #machinelearning
event: Chrome Dev Summit 2020; re_ty: Publish; product: Chrome - General; fullname: Jason Mayes;

[@JasonMayes]

Thanks for viewing the talk, if you have any further questions about TensorFlow.js come check out our booth at Chrome Dev Summit, or you can add me over on Twitter @jason_mayes / LinkedIn and ask me there!

[@slavko321]

Great presentation, thank you!

[@JasonMayes]

Thank you for the kind words

[@carpioerwinjohnt]

awesome future for javascript, the web and ai/ml

[@antongeorgiev1089]

A very pleasant and accessible explanation of the basics, thank you.

[@jsinnaour]

Thats awesome, ML + AI = life savers.

[@ErnestoSun]

The clients are the new servers. This feels like distributed web finally...

[@ahny1340]

ោ

[@ahny1340]

ោ

[@ahny1340]

ោ

[@ahny1340]

ោោ

[@felixinit]

Uff!... A lot of material to learn AI applying TFSJ in Javascript Web Browser!... Thanks!

[@JasonMayes]

If you have questions feel free to reach out on linked in or Twitter

[@xmlbk]

It's so cool! Thanks for sharing.

[@MarkLangerGER]

Fantastic video for a newcomer. Thank you very much, Jason! 💪🏼

[@JasonMayes]

Most welcome. If you have questions feel free to ask here or reach out to me on Twitter @jason_mayes or LinkedIn!

[@abhinavsirohi]

AI explained well 🙏

[@sofianezerrouki]

great technologies ...

[@jonasex3001]

Very very cool stuff. Great presentation.

[@mohitsingh5752]

Really simple and encouraging video .... Thank you Jason👍

[@jasonmayes_goog]

You are most welcome!

[@indylawi5021]

I've been doing ML with Python on the side. It is awesome that TF is available to add ML to Web dev with Javascript. Thank you for the info and demo.

[@mehdijarraya3196]

Awesome video, thank you soo much.

[@GabrielDalposso]

Coincidently I was recently searching for a way to do ML on the client side. Glad to see Google hearing my prayers 🙌 Thanks to everyone in the team!

[@JasonMayes]

Do let us know what you make if you end up using TensorFlow.js - we would love to see. Use the hashtag!

[@DTheHAge]

So all deep learnings are machine learnings, but not every machine learning is deep learning? Just kidding, very nice explaination. I don't know why I watched it, but I enjoyed it. It's very important that people realize that their phone and browser is already using multiple AIs and that we have the devices to calculate our Ais on our phones. This is in case of data protection a nice option. Imagine the AI that sorts the your mails for you (like in Gmail) would run on your device only. With this you could encrypt the model client-side and upload it for synchronization only. Or customized Ads. Just give the user a list of ads in different categories and let the AI locally calculate what the best ad for the user is.

[@JasonMayes]

Indeed people often assume JS is just a toy but it's actually very powerful and has matured a lot since the early days. Glad you enjoyed! Happy hacking!

[@marjanralevski3058]

I was dreaming about this in my college days. Combining web dev with ML. I guess dreams do come true. Awesome!

[@otorishingen8600]

lol - me too .. the funniest thing is that javascript just wont go away XD

[@JasonMayes]

JS is here to stay and I'm excited to see what you make! Do share if you do!

[@marjanralevski3058]

@@JasonMayes sure thing. And moreover, I couldn't agree more about JS.

[@9kopb]

🔥awesome, thanks! 🏄

[@kuperrr6776]

Awesome Job! thanks for the information!

[@JasonMayes]

Most welcome. Glad you enjoyed the presentation! Look forward to seeing what you create!

[@corky2812]

is there a tutorial link to get the face mesh with webgl cloud?

[@JasonMayes]

Not sure quite what you referring to but if you want the facemesh demo the details are all on the tensorflowjs website under models - click on facemesh and you will find the demo i showed on the github repo readme

[@victorpinasarnault9135]

I Love it!

[@andreadoimo3468]

AWESOME! best ML overview ever.

[@JasonMayes]

Feel free to share with others who may enjoy and you can find me on Twitter / linked in if any further questions on tfjs

[@natsaliya]

It's very inspiring!

[@JasonMayes]

Glad you enjoyed! If you have any questions reach out on Twitter @jason_mayes for a faster reply. Happy hacking!

[@rashadali9923]

Amazing Video!!! Thanks a lot

[@pulok9909]

Very much informative

[@JasonMayes]

Glad you enjoyed! Let us know if you make stuff!

[@_umarahmed]

Awesome! #TensorNinjas !

[@JasonMayes]

Glad you enjoyed!

[@ninjatwentyfive]

didn't knew that tf.js can do this much!!

[@JasonMayes]

Hope to see what you make in the future!

[@TopicalAuthority]

Fantastic video, thanks!

[@masteringreactnative]

Amazing! Can't wait to start learning

[@JasonMayes]

Glad you enjoyed! If you have any questions reach out on Twitter @jason_mayes for a faster reply. Happy hacking!

[@TomHermans]

Great video Jason. Very informative and inspiring

[@JasonMayes]

Glad it was useful! What else would you like to know about in the future in this space?

[@TomHermans]

@@JasonMayes practical projects. Setup, idea, training a model, applying it in a real project. With code and stuff. Just like a NN i'm very much a learner by example 😁

[@JasonMayes]

@@TomHermans thanks for letting me know. We have some codelabs you may enjoy in the meantime before videos are made. Check Google codelabs and search TensorFlow JS

[@eliceoreyes4910]

@@JasonMayes son muy buenos los proyectos

[@linfengwang9640]

awesome, it seems a lot of using sicario with tfjs

[@ralphlagos4210]

What’s the difference from using this and accessing an api that does tensorflow already like let’s say in python?

[@JasonMayes]

See towards end of talk where I talk specific advantages of running client side in browser such as privacy, lower cost, faster inference times (no server side call so no round trip time to server and back again etc)

[@ralphlagos4210]

@@JasonMayes Amazing, thanks Jason!

[@politicaltroll8920]

I didn’t hear a word I was hypnotized by the microphone. 🌈 🎙

[@JasonMayes]

RGB makes the ML run faster. :-)

[@TesterAnimal1]

The superpowers that we were promised in the sci-fi movies. Yep. Like 1984…

[@hanesmitter1469]

Uh! Now I know how superman shoots those laser beams from his eyes, after years of being dumb🤠

[@eliceoreyes4910]

Exelente más innobado y fácil de manejar

[@btsreeler4186]

Nice

[@franklemanschik_de]

I liked it a lot good starter material well done

[@JasonMayes]

Thank you Frank!

[@karamturki6597]

nice one , thanks

[@JasonMayes]

You are most welcome happy hacking!

[@wzafar77]

I learned if its above the line its an orange and if its below its an apple... damn... so much to learn...

[@JasonMayes]

Simple beginnings from which everything else evolves from.

[@naturegoggle]

thanks for this awesome stuff but Ironman doesn't shoot lasers from his eyes/mouth. :D

[@JasonMayes]

Whoops! Wrong superhero my bad. Maybe we can do from hand using handpose instead :-)

[@PeterMoueza]

Tensorflow.js vs Lite A timeline : Artificial Intelligence vs ML vs deep learning 10:40 ubiquity 10:46 transfert learning ML transfert 10:54 Facebook pose AR 11:20 pre trained 12:02 automatic scroll 12:36 detection vs recognition ... photogrammetry like ... examples 18:38 transfert learning 21:05 Auto ML 24:34 architecture 26:28 WebXR 26:53 resources 27:31 book 27:43 Codelabs 28:02 Community use cases 29:01 author

[@komakaze1]

Teachable Machine sounds like it might be easy and fun to play around with to learn and prototype machine learning!

[@Textras]

Evil laughing (at TF potential) muahaha

[@JasonMayes]

Do tag us with #madewithtfjs if you make something cool!

[@DaleReeves232]

iron man lasers lol

[@JasonMayes]

My bad. I should have said the baby from "The Boys" who actually shoots lasses from eyes

[@KT-ut9zg]

Public demo for the qna (github.com/tensorflow/tfjs-models/tree/master/qna) doesn't work: JS error: tfjs:17 Uncaught (in promise) Error: 'multiply' not yet implemented or not found in the registry. This kernel may not be supported by the tfjs backend you have chosen

[@JasonMayes]

Can you submit a bug report. It is possible its because its referencing latest version of TFJS that has breaking changed. Changing to fixed version for the library import eg 2.0.0 may make this go away.

[@jasonmayes_goog]

I believe this should now be fixed.

[@ibnusufajar1763]

IooJoo iini adalah untukmu kita ooooo

[@cleonunes6059]

Igor vai

[@bunny_rabbit5753]

Hey guys listen a joke 😁 "Thank you for leaning machine learning 😁😁😁😂😂😂😂😂😂"

[@ironmonkey4411]

Tensor floww?? no thanks

[@moonaka2]

why not

[@ironmonkey4411]

try an alternative, you will know

[@moonaka2]

@@ironmonkey4411 I'm new to machine learning so i was genuinely curious, Could you tell me why you think tensor flow is bad, and what alternatives do you think are better and why?

[@ironmonkey4411]

take a simple exercise and develop it with more than one tool, for example tensorflow, pytorch, keras or any other, in my opinion tensorflow makes things very difficult for you in many ways, is powerful but.. it is hard work.

[@moonaka2]

@@ironmonkey4411 Which one produces the highest quality models on average?

[@VishnuSunilkumar]

Nice