Jupyter Notebooks, Amazing (alternative to IDE) Tool or Satan's Demon Spawn: -It's an intellectual dance off.

 It turns out scientists and engineers like a good intellectual dance off as much as anyone else, so here we are.  It's 3 am and the back door to some warehouse in SOMA was left unlocked, so that we could all gather in the basement and watch an intellectual knock em, sock em, kick em when they're down, good old fashioned dance off.  David Bowie, were are you when we need you to judge these events?

In the corner of all that is good and decent in the universe, we have Jeremy Howard, one of the founders of fastai, that guy that seems to regularly show up in the first lecture in each HTC 'Getting Started with Deep Learning' course (remember, each HTC lesson has more than one lecture in it, that's why it's the HTC course, you get extra bonus material in the HTC course.  So he's that intellectual gladiator in the first set of weekly lectures we could all learn a new thing or 2 about programming from, at least that's the conclusion i've come to.

And Jupyter Notebooks are amazing beasts. As are things like Wolfram Mathematica Notebooks they seem to have evolved from (or are second cousins to).  The encapsulate many modern advancements in the whole notion of what a document can even mean in the world we fid ourselves living in today.  

Why shouldn't we care enough about the documentation of our code to make it's construction an integral (and also a beautifully well rendered) part of that code, included as a part of the Notebook entry describing the code.  Viewing code development through the lens of modern desktop publishing, as opposed to viewing it through the lens of a VT100 video terminal.

But enough intellectual banter, this is an intellectual dance off.  Let Jeremy's presentation begin.


And the first dance moves have now been thrown down.  In favor of the joyous wonder of Jupyter Notebooks.  

In response, Joel Grus (of the Allen Institute for Artificial Intelligence) will try and rock the house with his confession that he does not like Jupyter Notebooks.  And his subsequent presentation of such below.


Due to his untimely exit from this mortal plane, David Bowie will not be able to judge today's intellectual dance off competition.  But using a Ouija board, David has made arrangements for Iggy Pop to take his place, with some assistance by Josh Holme.  So they will be declaring the winner of today's intellectual dance off.  

Jupyter Notebooks, better than ice cream for keeping our brains focused, perhaps acting like an intellectual lens to organize our dis-ordered thoughts into neatly formatted text, that can be output as pdf, or html, or runnable fastai python scripts, filled with the vital chi energy apparently needed for developing optimal deep learning programming chops.

Or are they like that delicious apple that snake oil salesman snake convinced Eve to take a bit out of in that food court garden cafeteria, an act so scandalous the gossip about it never ceased to be discussed.

Comments

Post a Comment

Popular posts from this blog

CycleGAN: a GAN architecture for learning unpaired image to image transformations

Image
 Like the Doublemint Twins touting the joys of Doublemint gum, 2 GANs are surely better than 1 GAN.  Especially if we package them together inside of one meta GAN module.  And this is exactly what the CycleGAN architecture does. Have you ever harbored dark secrets of turning a horse into a zebra?  The CycleGAN was developed to do just that. Learn how to turn a horse into a zebra.  And more. Now right away you can notice a difference between the image to image transformation GAN architectures we've been discussing over the last few posts.  Those last few posts described systems that learn from a database of matched input-output image pairs. And if your goal is to turn an edge representation into a nicely filled in continuous tone image, it's easy to build your database of matched input-output image pairs that your GAN system can then learn off of.  Take a continuous tone image (which will be the output of the database pair entry), then run it through an edge detector algorithm. 
Read more

Pix2Pix: a GAN architecture for image to image transformation

Image
 I thought following up yesterday's TraVelGAN post with a Pix2Pix GAN post would be useful to compare what is going on in the 2 architectures.  Two different approaches to the same problem. I stole this Pix2Pix Overview slide below from an excellent deeplearning.ai GAN course (note that they borrowed it from the original paper) because it gives you a good feel for what is going on inside of the Pix2Pix architecture.    Note how the Generator part is very much like an auto-encoder architecture, but rebuilt using the U-Net architecture features (based on skip-connections) that fastai has been discussing in their courses for several years before it became more widely known to the deep learning community at large  (and which originally came from an obscure medical image segmentation paper) . So the Generator in this Pix2Pix GAN is really pretty sophisticated, consisting of a whole image to image auto-encoder network with U-Net skip connections to generate better image quality at highe
Read more

Smart Fabrics

Image
"What makes smart fabrics revolutionary is that they have the ability to do many things that traditional fabrics cannot, including communicate, transform, conduct energy and even grow."  This is a quote from Rebeccah Pailes-Friedman, who is a professor at Pratt with a research focus on developing wearable technology and smart textiles. Enhancing textiles can be viewed through different lens.  Enhancements can be aesthetic, or performance boosting, or durability boosting. 
Read more