I am running a research project into using evolution and genetic algorithms to create simulated nervous systems. There are technical challenges as well as philosophical questions. I have been finding some interesting parallels to real world evolution (what I'm seeing as an amateur and naive biologist). I am a professional engineer.
I would like to get reaction and feedback from this group. Two installments have been produced.
Episode 1 - A bit long but describes the technical underpinnings.
Episode 2 - Shorter, more animated as a swimming creature is evolved
I am currently adding learned behaviors.
I have a larger goal of discussing how to create a platform to support this effort.
Edited by Admin, : Put YouTube videos directory into message.
I at first assumed you were a college student, but the sophistication of what you were presenting (as well as the quality of the presentation itself) was beyond that, so I sought out your LinkedIn profile and now it makes more sense, but I haven't watched enough yet to understand why you're doing this, since there's already so much active research into neural nets and genetic algorithms.
How much help do you need? How much posting around on discussion boards are you doing? What tools are you using for online teamwork?
Abstract: Mathematical computer models of two ancient and famous genetic networks act early in embryos of many different species to determine the body plan. Models revealed these networks to be astonishingly robust, despite their 'unintelligent design.' This examines the use of mathematical models to shed light on how biological, pattern-forming gene networks operate and how thoughtless, haphazard, non-design produces networks whose robustness seems inspired, begging the question what else unintelligent non-design might be capable of.
Religious belief does not constitute scientific evidence, nor does it convey scientific knowledge.
Belief gets in the way of learning--Robert A. Heinlein
In the name of diversity, college student demands to be kept in ignorance of the culture that made diversity a value--StultisTheFool
It's not what we don't know that hurts, it's what we know that ain't so--Will Rogers
If I am entitled to something, someone else is obliged to pay--Jerry Pournelle
If a religion's teachings are true, then it should have nothing to fear from science...--dwise1
"Multiculturalism" demands that the US be tolerant of everything except its own past, culture, traditions, and identity.
Liberals claim to want to give a hearing to other views, but then are shocked and offended to discover that there are other points of view--William F. Buckley Jr.
Thanks for the reply. I see you also a New Englander (I'm from Ipswich).
This has been an interest and fascination for me for quite sometime. I decided to dust off some old skills see what I can do. I haven't coded in about 15 years, hence my choice of old school tools (c++ etc.).
You asked a great question, why.
There is a lot of interesting things going on right now regarding AI, neural nets, machine learning and genetic algorithms. I worked in Kendall Square Cambridge in the mid 80's next to what was called AI Alley. Every thing that was talked about then is happening now.
Neural nets and what I'll call mathematical machine learning has jumped the fence into a commercialization and industrialization phase.
The use of evolution (modeled like biological evolution) and using genetic algorithms as a tool is in a very interesting experimentation phase. Some have applied GA to specific problems (wind turbine optimization) and building simulated neural nets that do eye catching and whiz bang things (play Mario etc). MathLab now includes GA support tools.
I believe the time is right for a more industrial approach. Some of the hallmarks will be:
1. Consistent platform 2. A set of standard interfaces and methodologies 3. The ability to base work on past work (this was a key point to my first experiment). We are specialized fish. We carry in our brains, older versions that still have use. A platform needs to optimize in this dimension 4. Create an environment for multiple and disparate contributions
Happy to talk more about this in this forum, in direct communication or over coffee at the NH border. Feel free to reach my at my email, linkedin or smoke signal.