Superpowers with TensorFlow.js (TF Fall 2020 Updates)

[@arijit_student]

Super awesome! Thanks @Jason Mayes! 🤩

[@JasonMayes]

If anyone has any questions after watching feel free to drop me a message on twitter / linked in / here.

[@chrisminnoy3637]

Great video! Thanks. Now I know how to keep me busy during this Christmas holiday.

[@dathirdsugian2689]

Super awesome 🔥🔥🔥 Thanks for this talk 💯💯💯

[@neeban1339]

So coooool! Will surely try this for e-commerce solutions.

[@MrVicky47]

Please make a video on running a pre-trained pose models on mobile devices.

[@nmstoker]

Love the video. Would be great if you did a similar one for NLP applications of tensorflow.js

[@JasonMayes]

Check out my codelab for spam filtering and such!

[@rajumahto4537]

@@JasonMayes od

[@mjaathik]

Damn Super Informative. Thanks👍

[@JasonMayes]

Glad it was of use!

[@quantum_field]

Enjoying these, thanks!

[@JasonMayes]

Glad you like!

[@huzaifakhan_771]

7:55 Iron man doesn't shoot out lasers from his eyes or mouth. It's more like Cyclops

[@harshrai6259]

I was hoping that someone would have commented or I will : ) and also Cyclopse has one eye its more like Superman : )

[@piotr780]

Is tensorflow.js popular at all? I don't see any mentions in job offers

[@luis96xd]

Excelent video! Tensorflow .JS is so cool😄😄 I have a doubt: I'm making a regression model and it receives an array [36,1] for Xs and Ys a relationships between X and Y Then I created a dense layer units:1, inputShape: [1] But when I use model.predict(tf.tensor2d([20], [1,1])) it gives me a NaN value What should I do? 😨

[@henkhbit5748]

Super exiting stuff

[@thegalhorowitz]

hey Jasonnn 🙌🏻🙂🍻

[@JasonMayes]

Hello there!

[@soja2897]

Nice and cool 😎

[@username42]

what about 1D tabular data classification ?

[@JasonMayes]

You can indeed do this too just like regular TF but in JS!

[@username42]

@@JasonMayes any source links? since i could not find it yet usually it is for 2d image data but nothing for 1d tabular tf js yet seen :/

[@JasonMayes]

@@username42 So you want like a simple linear regression? Typically a table is considered 2D data as you have rows and cols even if just 1 row.

[@JasonMayes]

Check this out: livebook.manning.com/book/deep-learning-with-javascript/chapter-6 gist.github.com/learncodeacademy/a96d80a29538c7625652493c2407b6be

[@username42]

@@JasonMayes not a simple linear regression at all, i have a tabular data set which as thousands of variables and hunderds of samples from a sensory data set obtained from instruments. and trying to classify my samples accordinly based on the variables i have but not even ann or deep learning solves it more than 80% sometimes and not even sees such a nice classification models for tabular data sets, always for image data set which is 3d since rgb colors have 3 different columns and cnn uses 2d for it. but for instance for my such data i need 1d cnn and do classification, i have already tried xgboost, catboost and etc, but not working very well. also tried other standart supervised techniques like svm,random forest and etc. and also i cant use gan for such sensory data set which changes a lot and depends on the instrument so cant be use like image data augmentation techniqeus as we do it in gans and etc.

[@anclaudys]

Hey, how can I get this to work with a node back-end or a GPU back-end? It seems like all of the tutorials out there are about using this with the browser link....

[@piotr780]

It's for client side... Do you want access to user GPU - forget...

[@ricardoferreirades]

This is powerful!

[@b09a321]

Nice tutorial

[@JasonMayes]

Thank you!

[@encoder10101]

Awesome!

[@henricastro4063]

awsome

[@JasonMayes]

Glad you enjoyed!

[@debkumarmishra1426]

TensorFlow.js community is not big as python.

[@JasonMayes]

If I do my job right one day it will be larger as 70% of engineers use JavaScript. Chrome also didnt used to be the most popular browser, but that changed, so keep an eye for the future as we are growing exponentially year on year in usage.

[@debkumarmishra1426]

@@JasonMayes yeah .. Will google launch TensorFlow Golang version ?