Profile cover photo
Profile photo
Artificial Brains
3,226 followers -
The quest to build sentient machines
The quest to build sentient machines

3,226 followers
About
Artificial Brains's posts

Post has attachment
World's 4th largest supercomputer simulates 1.73 billion neurons

Using the NEST software package the team succeeded in simulating a network of 1.73 billion neurons connected by 10.4 trillion synapses. To realize this feat, the program recruited 82,944 processors of the K computer in Japan. The process took 40 minutes to simulate 1 second of biological neural activity.

Press release: www.riken.jp/en/pr/press/2013/20130802_1/
Photo

Post has attachment
Monkey brain wiring diagram wins visualization challenge

The attached image shows the long-distance neural connections between major regions of the macaque monkey brain. It's a fairly large image, over 2,000 pixels high, so worth viewing on a full-size screen. Use the mouse wheel to zoom.

The visualization was created by +Dharmendra Modha's team at IBM as part of the DARPA SyNAPSE program. It is used to provide guidance for connecting thousands of neuromorphic chips into a brain-like network.

This week the image made the cover of sciencemag.org by winning first place in the illustration category of their 2012 Visualization Challenge.

Article: sciencemag.org/content/339/6119/512.full
Background: artificialbrains.com/darpa-synapse-program
Photo

Post has attachment
Human Brain Project wins €1 billion in funding

The Human Brain Project is one of two European flagship science projects to win €1 billion in funding over ten years. The official announcement isn't due until January 28, but Nature says the decision has already been reported by Spanish and Italian outlets. The other winning project is Graphene.

The Human Brain Project involves 80 institutions that will attempt to model the human brain and build a complete simulation of it within a supercomputer. Goals are to understand how the mind works, to find new treatments for brain disease, and to develop new kinds of computers. The project will start in September 2013 and full brain simulations are targeted for 2023.

- Nature article: goo.gl/IDfbn
- Website: humanbrainproject.eu
Photo

Post has attachment
Ray Kurzweil joins Google as a research director

Ray Kurzweil announced yesterday that he's joining Google as a research director. He starts this Monday and will work full time at the Google HQ in California. His work will focus on machine learning and language processing.

Last month Kuzweil published a book titled How to Create a Mind. In turn, Google's long-term aim is to build a machine that is AI complete. So it makes perfect sense that Kuzweil and Google should team up.

Brain simulation, or building a full-scale functional model of the human brain, requires huge amounts of computing power. Currently nobody has more compute power than Google. In our view it's very possible that Google will succeed in creating the world's first AGI.

Announcement on kurzweilai.net: goo.gl/5BKIb
Photo

Post has attachment
Spaun - world's largest functional brain model

Spaun is a simulated brain that contains 2.5 million neurons. This is far fewer than the 86 billion in the human brain, but enough to recognize lists of numbers, do simple arithmetic, and solve reasoning problems.

The system is biologically realistic in its simulation of spiking neurons and neurotransmitters. As a result, it reproduces many quirks of human behaviour, such as the tendency to remember items at the start and end of a list better than those in the middle. 

Spaun - the Semantic Pointer Architecture Unified Network - was created by +Chris Eliasmith and colleagues at the University of Waterloo in Canada.

Article published in Nature today: goo.gl/A02aX
Project homepage and videos: nengo.ca/build-a-brain/spaunvideos
Photo

Post has attachment
Blue Brain Project - Year Three video released today

The third part of +Noah Hutton's excellent documentary about the +Blue Brain Project was released today. This is a ten-part film, with one part being released every year. It follows the project team as they attempt to build a complete simulation of the human brain within a supercomputer.

Weblinks:
 - bluebrainfilm.com - all video parts
 - bluebrain.epfl.ch - project homepage
 - artificialbrains.com/blue-brain-project

A few project members are active on Google+. They include the following, if you'd like to circle them: +Marc-Oliver Gewaltig, +Melissa Cochrane, +Stefan Eilemann, +Kamila Markram

Post has attachment
Ray Kurzweil's new book released today:
How to Create a Mind

This is a probably a must-read for anyone interested in how artificial general intelligence will be achieved. His basic premise is that the brain contains no hidden secrets and that the creation of sentient machines will follow from re-engineering the brain in non-biological form.

Kurzweil estimates that the human neocortex contains 300 million pattern processors linked horizontally and vertically. He claims it is these processors, rather than the neurons of which they are composed, that are the fundamental units of the brain. Near-future computer technology could build a synthetic brain containing well beyond "a mere 300 million" processors, and instead as many as a billion or a trillion.

Website: howtocreateamind.com
Photo

Post has attachment
Brain simulations and the new TOP500 supercomputer list

Twice a year a list of world's the top 500 supercomputers is released. Today the list was updated and the new Titan supercomputer in the US took top slot at 17.59 petaflops. Rising to 5th place, with 4.141 petaflops, is the JuQUEEN machine in Jülich, Germany.

If the +Human Brain Project is successful in securing EU funding in January 2013 then the brain simulations will be run on JuQUEEN. This machine is now around 100x more powerful than the supercomputer currently used by the +Blue Brain Project in Lausanne, Switzerland.

JuQUEEN (pictured here) should be powerful enough to simulate all 100 million neurons of a mouse brain.

Details: www.top500.org/lists/2012/11/ and humanbrainproject.eu
Photo

Post has attachment
New video about the +Human Brain Project

Yesterday the HBP released a great new video giving an overview of the project. It describes how scientists from over 80 institutions across Europe are attempting to build a computer simulation of the complete human brain. Goals are to understand how the mind works, to find new treatments for brain disease, and to develop new kinds of computers.

Last month a proposal was submitted to the European Union requesting €1 billion in funding over ten years. The EU will make a decision in January 2013 on whether to award the grant. If funding is given, the project will start in September 2013 and full brain simulations are expected by 2023.

- www.humanbrainproject.eu
- www.artificialbrains.com/blue-brain-project
- video found via +Mark Porritt

Post has attachment
Human Brain Project
Status update, October 2012

Last week the leaders of 90 groups involved in the HBP convened in Lausanne for a summit meeting. This was a private invite-only event. By pure coincidence I happened to be in Lausanne on Tuesday when I saw a sign saying "HBP summit this way". So I followed the arrow into the Forum Rolex conference center and they were kind enough to let me sit in for a few hours.

By way of a reminder, the Human Brain Project (HBP) is a proposed ten-year project to understand and reverse engineer the human brain. The project is competing for a €1 billion grant from the European Union. The HBP is a worldwide consortium, to be based in Europe, and led from EPFL in Lausanne, Switzerland. The aim is to bring together hundreds of neuroscientists, computer scientists, roboticists, and neuromorphic engineers in a concerted effort to finally understand the brain.

Current status

The current status is that finishing touches are being put to the proposal document. A 108-page draft has already been available online since July (goo.gl/eDZKk). The final version must be submitted by the deadline of 5pm on Tuesday 23rd October 2012. During November the proposal, along with five other competing projects, will be studied by scientific reviewers. It's not known who the reviewers will be, but they'll be appointed by the EU. In the second week of December the HBP leaders will attend a half-day hearing with the EU. I think this is to be in Brussels.

After the reviews and hearing, the reviewers will have a consensus meeting and the six competing projects will be marked according to their scientific merit. Together the reviewers will then write a report and send it to the national referees from EU member states.

In principle, the one or two projects that score best in the scientific evaluation will be the ones that win funding. If two or more projects are tied or nearly tied, then the choice of project will become political. This is expected to happen. In fact, it was mentioned at the HBP summit that the decision would likely be around 70% political.

Final decision date

The final decision is officially due to be made in mid-January. At the meeting last week, however, it was said that the decision would probably be delayed. The wheels of bureaucracy turn slowly in Brussels and it seems unlikely that a decision involving as much as €2 billion could be made so quickly. Nevertheless, a decision is expected sometime between January and March. Most likely in February 2013.

Once the decision has been made there will then be a six-month "negotiation phase". I'm not sure of the details, but this probably involves agreeing on the finer points of exactly how much money is awarded and to whom and when. Once this is completed the project can then get underway, probably around September 2013 and continuing through until September 2023.

From the general comments made at the summit last week, the HBP leaders seem very confident they'll get the funding. But it is not yet a foregone conclusion. They will be working hard these next two weeks to get the wording of the proposal correct and to make sure that the value of the project is conveyed with maximum impact.

€1B is not much given the scale and importance of the problem

During the course of the meeting some comments were made that a billion Euros sounds like a lot of money to scientists. But in reality, given the scale and importance of understanding the human brain, it's actually not that much. Indeed, when compared to the amount spent by governments on other things, such as bank bailouts and war, then one billion is a drop in the ocean. It was mentioned that in science there are two big fundamental questions: 1) understanding the universe, 2) understanding the brain. The HBP could go a very long way to answering the second one. The long-term payoff from this project is potentially huge. It should be given proper funding instead of having to survive on crumbs.

Comparison with the Human Genome Project

At the meeting some comparisons were drawn with the Human Genome Project. If I understood correctly, there was lots of controversy in the late 1980s about that endeavour. Apparently many people doubted the value it might bring. Since its successful completion, however, the field of bioinformatics has exploded. Nowadays many thousands of scientists all over the world have little trouble getting funding for genome research. The genome project has had knock-on effects that have brought a lot more funding into the field. The HBP could be very similar. By creating a successful and integrated platform for brain research, the groundwork will be laid to attract lots more research funding in the future.

Official website: humanbrainproject.eu
Attached photo from the HBP summit, 4th October 2012
This post written by +James Pearn 
--
Photo
Wait while more posts are being loaded