[HacktionLab] workshop report (Neuroscience & Computing) for Barn Camp wiki
m.reinsborough at qub.ac.uk
Mon Jun 22 16:02:43 UTC 2015
Hi everyone, great weekend! It was good to see so many new and old faces.
..below (and attached as a text file) is text for the BARN CAMP wiki on workshop reports. I don't have a wiki page access so could someone who does have this put the text up on the wiki in the appropriate location.
Report on workshop/discussion
Neuroscience & Computing
315pm June 20th 2015 at Barn Camp by Hacktionlab, Wye Valley, UK
This discussion began with 1) a round of introductions, then there was 2) a brief presentation on Neuroscience & Future Computing and finally 3) discussion, both at the end and interspersed through the presentation. A second briefer presentation on Artificial Intelligence happened during the discussion.
here is a few (not all) of the things people said
"I do data analysis. I'm interested in neural networks. Although that's not the type of data analysis that I do but I've heard of it"
"I'm a programmer. I do a lot of 'crypto'[graphy]"
"I watched that film ex_machina recently"
"I do user experience on a website design team"
"I used to be a paleontologist, now I'm a hippy"
"I do lo tech"
People may not be aware of the connection between neuroscience and computer science. The exchange between developments in computer science and developments in neuroscience is two way: Firstly, neuroscience is now being done more often with computational techniques. Secondly new techniques for computer technology are being sought the principles of the brain, that is by thinking about the brain as if it were a computer and reverse engineering the brain to make computers that use less power, think in alternative ways, and do much more parrallel processing. As has previously happened many times, people often turn to copying, privatising, borrowing or stealing from the natural world in order to jump start the economy.
There are now various initiatives in Big Science that are doing work on the brain. Big science initiatives are a mixture of science + industrial policy + PR & media. Sometimes they include some 'ethics and society' initiatives to discuss the social impacts of new technologies developed from the initiative. Sometimes they don't. Big Science initiatives are usually modelled after the original and most famous Big Science initiative which was the Human Genome Project to map the genetics of a human being. While there are various national initiatives on brain science the two largest are the Obama BRAIN initiative https://www.whitehouse.gov/share/brain-initiative and the EU funded Human Brain Project http://www.humanbrainproject.eu . The US based Brain initiative is partially funded by military sources whereas the EU initiative does not accept any military funding.
The EU project is to build infrastructure platforms to do neuroscience. It's broken into three areas: Future Medicine, Future Neuroscience, and Future Computing & Robotics. Although this is called a neuroscience initiative the budget for it came from the EU computer science research budget to support research and economic competitiveness in the European computer industry. This led to some public controversy at the beginning of the project.
Future Medicine: USing Big data and patient records, the medical initiative wants to link mental health problems (which are currently very difficult to understand causes for) to 'biomarkers.' Biomarkers are something measurable and biological that can be used to help identify and diagnose a medical problem in an individual person. The new EU data protection legislation may affect the ability to use big data for medical research & there is a proposed exception for medical research.
[Discussion: There was a discussion about where people saw mental health problems as coming from. Most people said mental health problems were a social issue, not biological. One person said they thought some people had a predispositions to have mental health problems that were triggered by difficulties in social environment. Most people in the group saw the brain as changing over time from birth on and mental health issues were developed. Question: Is this represented in the research? Answer: No current research is only attempting to develop a single picture of the brain, like a snap shot. It is not trying to find developmental pathways of mental illness. It sees mental illness as a problem in the brain and not as a social issue. There was also discussion about the recent attempts by government to make everyone's medical records available for research purposes (even to potentially to private industry) which failed because of public discontent http://www.theguardian.com/commentisfree/2014/feb/28/care-data-is-in-chaos ]
Future Neuroscience: The neuroscience initiative is attempting to build simulation platforms as an infrastructure not just for neuroscience but for other types of large scale computer simulation work across Europe. The project is attempting to build or join up a computer modelling community for brain research.
Future Computing and Robotics: By reverse engineering the brain technologists hope to develop new computing technologies and High Performance Computing. Neuromorphic computing is one way to do this- this is computing using principles similiar to the tangled circuits of how neurons stimulate one another within the brain. There are two types of neuromorphic computing- one is emulating the brain through software, the other is making hardware that acts like circuits of neurons. These are called neuromorphic chips. These type of chips are better at recognizing patterns than a normal logic gate computer chip. Therefore many new 'algorithms' are coming out of neuroscience. Algorithms are increasingly being used by many technical systems that affect our lives. http://www.newscientist.com/article/mg22530073.500-living-with-the-algorithms-that-run-our-lives.html#.VYgcIU1waM8 The future computing initiative is also commissioning a super computer larger than anything now existing to simulate a full human brain. All of the half dozen or so companies that make supercomputers are also involved in working with the military for some of their contracts. Robotics may also be affected by new research in neuroscience. Theories of 'cognitive architechture' (how the brain is structured to make different types of decisions) are being used to develop robots. Robots are said to be needed for work in the 4 D's (dangerous, dull, dirty, or delicate) which humans couldn't or wouldn't want to do. Because brain like machines are better at recognizing patterns robotics and robotic software systems might be able to do new kinds of work in the future particularly if researchers can learn from or copy visual systems neuroscience. Many more professional jobs (journalist, legal assistant, etc.) might be able to be done by robots/robotic systems in the near future. It's important to remember that even now robots can't work alone. There always has to be human workers to watch the machines and fix problems in a factory full of robots, for example an automobile manufacturing center. The neuroscience & computing overlap is potentially also part of developing 'artificial intelligence (A.I.)' systems. A.I. has many corporate and banking sector uses, as well as military uses as well, such as military drones. It might also have medical or legal system uses. A.I. isn't actually like in sci-fi movies with thinking computers (HAL from 2001: A space odyssey is the most famous) but are perhaps better described as 'brain like machines' which are often very good at pattern recognition or matching. Although we don't notice it A.I. is also part of many everyday things we use like Google search engines and maps. Google is involved with researching A.I., robotics, self-driving cars, and they may also be involved in military research contracts or supporting government intelligence work.
[Discussion: Where do you think robots or A.I. systems will be used? Various answers from group- As servants, to reduce human work on processes, in transportation, agriculture, sex, legal profession, medical profession. Discussion about will the machines take over: one participant suggests they already have been, 200 years ago during the industrial revolution they started to take working class jobs, now they will start to take middle class jobs, but the 'Skynet' type fantasy is only ruling class people worrying about their jobs being taken over (boss everyone around and kill anyone who disagrees). People talked about the most famous A.I. project WATSON which won jeopardy, about memristers (?), about the video on utube 'Humans need not apply', ]
Second Presention (slightly shorter) by another participant of group on Artificial Intelligence (A.I.) and various companies involved in this: for example DeepMind (a computer that plays 2Dimensional video games from the 80s (discussion: does it only listen to music from the 80s as well?), DeepMind now bought up by Google, the work of Ray Kurzweil (famous for advocating the 'singularity', a sort of rapture for geeks) on optical character recognition, WATSON for IBM, zabbawhere- ultrahal computer based bot. THe focus of presentation was what is A.I. and why do we want it anyway?
[more discussion - is there now a breakfast companion that will talk to you while you eat your cereal? (?), viruses and botnets are designed to hide themselves and adapt to their environment so perhaps these are the real A.I.'s developing. Interest in theories of a holgraphic universe from physics; and interest in simulation. "we should be developing A.I. to learn in it's own environment which is a digital environment." "What's a brain anyway? Too much emphasis on this one bit of the body" "The brain is a big memory store. In my family we learned a lot from my mother's mental health problems. She had a breakdown in the barriers that stop information coming in. Her brain didn't stop all these things coming into her experience" More discussion about 'Affective computing.' This is when computers can recognize or simulate human emotions, potentially being able to influence human beings. One participant gave a somewhat linked example where marketing and voter preference information was used to encourage people to vote on green issues. Different voters were presented the issue in quite different ways. Another participant argued that there was a limit to how much people could be manipulated before they withdraw their consent.]
-------------- next part --------------
An HTML attachment was scrubbed...
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
More information about the HacktionLab