Neurdon has moved!

The new address is www.neurdon.com

see you there!

A Real Test for Object Recognition

Humans are remarkably good at identifying the same face across illuminations, positions, deformations, and depths. The same face can even be identified through fences, glass, and water. The possible number of contexts for a face to appear in are infinite, yet we can identify it instantaneously. For whatever reason, we are really good at identifying objects, but researchers have struggled to make computers even semi-competent at it. One of the more valiant efforts is Yann LeCun’s use of convolutional nets, but its primary successes are in controlled situations. Any reasonable person in the field would agree that any human can wipe the floor with even the best algorithm running on the best supercomputer (programmed by the best programmer in the best department in the best state in the best country!). So what gives?

IBM Seeks to Build the Computer of the Future Based on Insights from the Brain

In december 2008, a video post has been published on Abovetopsecret.com  with the title “DARPA & IBM building a “global brain” “cognitive computer” for “monitoring people”. In this video, the leader of the IBM SyNAPSE project, Dharmendra Modha, talks about SyNAPSE.

This is an excerpt from the video:

“Hello World” on Memristive Nanodevices

SyNAPSE is not a project DARPA undertook lightly. Many attempts at large-scale neuromorphic engineering have been made in the past. None met their goals. As such, SyNAPSE owes its existence to a number of recent game-changing developments. From HP Labs, the discovery of the memristor was one such keystone innovation. It took Greg Snider’s 2007 work in Nanotechnology, however, to establish memristors as a viable platform for the implementation of self-organizing recurrent neural networks.

The Brainputer

After running through the Businessweek article posted by Max, I am equally excited and nervous. Anyone has to be excited over the prospect of a new computing paradigm, though honestly I’m not sure what that looks like yet. These sorts of articles claim that computers will look more like brains, which is all well and good, because brains tend to do dominate the “competition”, i.e. computers, at messy things like object recognition and speech recognition. Conversely (and obviously), computers tend to dominate tasks amenable to decomposition into easily formalizable sequential steps, e.g. chess or even eye surgery. So, maybe we know what Deep Blue looks like, but what on Earth would a computer expert in messy things, a messy computer if you’ll excuse the phrase, even look like? We all agree that computers stink at these messy things, and if they didn’t stink at them it would be a huge boon to, well, humankind. So let’s make the computers more like brains so they can do what brains do so well! But how do we make computers, both in terms of hardware and software, more like brains?

IBM and SyNAPSE on Dharmendra S Modha’s Cognitive Computing Blog

Dharmendra S Modha is the Principal Investigator in one of the three DARPA SyNAPSE grants, the one awarded to IBM. Modha is the Manager of the Cognitive Computing facility at IBM. Here is the full article from his blog.

SyNAPSE in BUSINESSWEEK

Making Computers Based on the Human Brain
How the biology of gray matter is having an increasing influence on computer design

HP and SyNAPSE

Link: http://www.eetimes.com/news/latest/showArticle.jhtml?articleID=212200673

Dr. Snider and his colleagues at HP have built an integrated hybrid circuit with both transistors and memristors. Memristor crossbars are a very promising technology that can ultimately lead to building very dense hybrid chips, several times denser than synapses in the human cortex. Also, memristors have shown the potential to mimic the learning functions of synapses in neural networks. Memristors will the key technology that HP and its academic partner, Boston University, will leverage in the SyNAPSE grant.

SyNAPSE on the NY Times

On November 20, 2008, the NY Times has published a short article entitled “Hunting for a Brainy Computer”. Steve Lohr interviews the leader of the IBM team. IBM’s Blue Gene has been used to simulate large-scale neural models (see the Blue Brain Project, led by Henry Markram). However, it is easy to mix supercomputers, IBM, and SyNAPSE in a big pot, thinking that they are the same. In reality, the Blue Gene is the example of how not to simulate the brain. This machine, as large as a room, whose power consumption is the same as the sum of the brains of a small city, can barely simulate a cortical column. As this article does not stress much (unlike other cited in this blog), the hardware problem will be solved (hopefully) by nanotechnologies, in particular by porting to nano the immense number of synapses that will link the millions of neurons implemented in the chip. No comment on “Dorothy looking for the Wizard of Oz” and “Want a really intelligent digital assistant”… It is worth mentioning that even with a chip twice the density and half the power consumption that the one SyNAPSE seeks to have in seven years available TODAY in the hands of the best modelers in the world, it is hard to think that we have the necessary modeling skills to implement that is suggested below.

SyNAPSE in IEEE Spectrum Online

HP memristor ...in perspective

IEEE Spectrum online. Again, IBM appears all over the news. One of the main misconceptions of SyNAPSE is that, imagining of course the 3 companies involved in SyNAPSE succeed, the resulting chip will automatically result in better “MRAPs, UAVs, Mars Rovers”. This is of course not true. A very dense neural chip is 1/2 of the story. The ingredient that SyNAPSE needs to succeed is having meaningful neural models implemented on the chip. And this is where the other 1/2 of the competition will lie in the long (7 years) program.