Profile cover photo
Profile photo
Stephan Gouws
246 followers
246 followers
About
Posts

Post has shared content
We are open-sourcing our internal deep learning framework!
Open Source Release of TensorFlow

I'm very excited to announce the open source release of the TensorFlow machine learning library that I and many others at Google have been building.

This blog post gives an overview:
  http://googleresearch.blogspot.com/2015/11/tensorflow-googles-latest-machine_9.html

The web site has a number of tutorials and documentation about the system:
  http://tensorflow.org

The source is on GitHub:
  https://github.com/tensorflow/tensorflow

You might also be interested in the whitepaper we've prepared that describes TensorFlow in more detail:
  http://tensorflow.org/whitepaper2015.pdf

We'd love to hear what people think.
Add a comment...

Post has attachment
As a novice when it comes to quantum computing, I really like how +Hartmut Neven  is able to convey these ideas using simple analogies without trivializing the content too much, or patronizing the reader. I just like his writing style in general.
Add a comment...

Post has attachment
Nice talk on the potential alienating effect of information personalisation by sites like Facebook and Google. I think what he is not considering is the normalising effect of social sharing of content: What was blocked to me was seen by a friend and retweeted, plus-oned or posted on his Facebook, thereby at least mitigating the "filter bubble" effect. http://www.ted.com/talks/eli_pariser_beware_online_filter_bubbles.html
Add a comment...

Post has attachment
I've shared this before, but this is worth reading again: Ariah Haghighi writes that most interesting real-world problems are not simply solved by thowing an off the shelf machine learning solution at it; knowledgeable insight into the problem, allowing us to decompose the problem space into several smaller, more manageable sub-problems (possibly using simpler ML techniques to solve each sub-problem) is the way to go. Conversely, most significant performance improvements are usually the result of this process, rather than finding "the ultimate" one-size fits all ML solution. http://radar.oreilly.com/2012/04/great-machine-learning-products.html
Add a comment...

Post has shared content
LOL :D
Add a comment...
Wait while more posts are being loaded