Profile

Cover photo
Deepak J. Nath
Lives in bangalore
2,535 followers|179,561 views
AboutPostsPhotosVideos

Stream

Deepak J. Nath
owner

State of the Art  - 
 
GPU real time tagging
 
Last week at the GPU Technology Conference, Nvidia CEO Jen-Hsung Huang gave a keynote speech in which he talked about deep learning and neural nets. He showed a real-time demo of our OverFeat image classifier. They let people upload pictures of dogs on Twitter and used our convolutional net demo to recognize the breed of the dogs.

GPU Technology Conference 2014: Machine Learning Demo (part 4) GTC

When I show this demo in my talks, I run it in real time and use a webcam to recognize objects around the room. I'm taking a risk every time I do this, but it makes the demo more interesting.
1
Add a comment...

Deepak J. Nath
owner

Conferences  - 
 
Unconference July 2014
 
I think I'll need to start pushing the Computational Intelligence Unconference again soon ( http://ciunconference.org/uk/2014/ ). Anyone wish to help with spreading the word?
(we've released a 50 more tickets too!)
1
Add a comment...

Deepak J. Nath
owner

State of the Art  - 
 
+Franco Cortese this is something of interest to you I believe.!
Living cells capable of emitting light and conducting electricity have been created in a lab, and MIT's researchers think this is just the beginning.
1
Add a comment...
 
Gravitational Waves!
 
Evidence of Cosmic Inflation and the Big Bang

#bigbang #physics  
1
Add a comment...
Have him in circles
2,535 people

Deepak J. Nath
owner

News  - 
 
Biomorphic ICs in the news
 
In his latest column in the New Yorker, my NYU colleague (and regular New Yorker columnist) Gary Marcus attempts to dampen the hype surrounding AI in recent press articles.

In particular, he takes issue with John Markoff's front-page article in the NYT about neuromorphic chips. I agree with many of Gary's comments. John's article is slightly misleading in that the chips that are mentioned in the article, and the type of technology they employ are very far from practical applications. 

Let's take the example of a typical convolutional net for image recognition, of the type that Google, Baidu and Facebook use. Such networks typically have around one billion connections, and 50 to 100 million parameters. Once they are trained, running them on an image takes a few milliseconds on a standard GPU card (e.g. a high-end gaming card that can be had for about $1000). The current breed of neuromorphic chips mentioned in the article are quite far from being able to run such networks. Some of the chips implement learning algorithms. But the algorithms they implement are simplistic, and are not the ones that actually work in practice (e.g. backprop+SGD). 

I'm not saying research on this topic should be stopped. But it's quite far from being practical, and it's not ready for prime time (certainly not ready for the front page of the NYT).

Furthermore, I do think that there is value in building specialized hardware for neural nets. I believe that there are embedded applications for which a compact, low-cost, low-power consumption chip that could run large pre-trained convnets could be very useful. We have done work in that direction in my lab at NYU (see http://www.neuflow.org/ ).

http://www.newyorker.com/online/blogs/elements/2014/01/the-new-york-times-artificial-intelligence-hype-machine.html
1
Add a comment...
 
Machine Learning
 
David Blei wins the ACM Infosys award for Topic Modeling.  http://awards.acm.org/infosys/  This is thrilling news for machine learning!  I'm also proud to have co-authored papers with 3 of the 8 ACM Infosys recipients so far. :)  http://awards.acm.org/infosys/year.cfm 
1
Add a comment...

Deepak J. Nath
owner

News  - 
 
Deep Learning
 
Today's edition of All Things Considered on NPR has a very nice segment on deep learning in which I am interviewed, and my dear colleague +Gary Marcus  comments on AI, deep learning and such.

Featured in the segment is the work of my just-graduated PhD student +Pierre Sermanet on recognizing cats from dogs (his winning entry in a recent Kaggle competition).

http://www.npr.org/blogs/alltechconsidered/2014/02/20/280232074/deep-learning-teaching-computers-to-tell-things-apart
1
Add a comment...

Deepak J. Nath
owner

State of the Art  - 
 
Thought extraction
 
Researchers Reconstruct Facial Images Locked in a Viewer’s Mind

Full article at http://goo.gl/YSSwG1.

Using only data from an fMRI scan, researchers led by a Yale University undergraduate have accurately reconstructed images of human faces as viewed by other people.

The research is in Neuroimage. (full access paywall)

Research: “Neural portraits of perception: Reconstructing face images from evoked brain activity” by Alan S. Cowen, Marvin M. Chun, and Brice A. Kuhl in Neuroimage. doi:10.1016/j.neuroimage.2014.03.018 In Press, Uncorrected Proof

Image: Researchers wanted to know whether it would be possible to reconstruct a human face from patterns of brain activity. The task was daunting, because faces are more similar to each other than buildings. Credit Alan Cowen.

#neuroscience   #neuroimaging  
2
2
R. Jacob Vogelstein's profile photoYasin Gülener's profile photo
Add a comment...

Deepak J. Nath
owner

Discussion  - 
 
Vicarious infused with 40 Million Dollars from Facebook.
1
Joshua Augustus Bacigalupi's profile photo
 
Really found this article informative and level headed.
Add a comment...
People
Have him in circles
2,535 people
Work
Occupation
Engineer
Basic Information
Gender
Male
Story
Tagline
You are better than you think
Introduction
Tech enthusiast; Interested in neuroscience, programming, the brain and artificial intelligence. I am looking for partners in my Biologically inspired cognitive architecture research.. Lets intersect our circles and hang out..! 


Bragging rights
Still have to earn it
Places
Map of the places this user has livedMap of the places this user has livedMap of the places this user has lived
Currently
bangalore
Previously
trivandrum
Links
Other profiles
Contributor to