Herself’s Artificial Intelligence

A Twitter friend working in artificial intelligence clued me into this cool open source project that’s just getting going.

Stephen Larson has a talk on Evolving AI: Lt. Data will be Born from Artificial Worms, you can watch the talk on YouTube.. There is a virtual worm project on Caltech

The general idea is to start with a digital form of a simple life form and evolve it into a general intelligence much as we have evolved from simple origins. Unlike the real world the computer can run through millions of evolutions in a very short time.

An open source project openworm has been started. You’ll find source code and demos there.

To bring you up to speed on worms you’ll want to dig through C. Elegans II an online textbook on C. elegans.

Also you’ll want to dig into WebGL which is used to create the 3D world demos.

So many claims, it’s difficult to sort the intelligent from the educated. But at some point one or more of these claims will be true.

FOR generations, the Avidians have been cloning themselves quietly in a box. They’re not perfect, but most of their mutations go unnoticed. Then something remarkable happens. One steps forward, and that changes everything. Tens of thousands of generations down the line, some of its descendents will evolve memory.

Avidians are not microbes, or sci-fi alien life forms. They are the digital offspring of Charles Ofria and colleagues at Michigan State University (MSU) in East Lansing. They “live” in a computer world called Avida, and replicate using strings of coded computer instructions instead of DNA. But in many ways they are similar to real life: they compete with each other for resources, replicate, mutate, and evolve. They – or things like them – might eventually evolve to become artificially intelligent life forms. read more Artificial life forms evolve basic intelligence

More information:
HyperNEAT
Avida-ED
Avida-ED user manual ( pdf )
Observing mutations on the genomes of evolving Avidians (pdf)
Harnessing Digital Evolution ( $$$ )
Digital Evolution with Avidian

Cuckoos have an aggressive reproduction strategy that involves the female laying her fertilised eggs in the nest of another species so that the surrogate parents unwittingly raise her brood. Sometimes the cuckoo’s egg in the nest is discovered and the surrogate parents throw it out or abandon the nest and start their own brood elsewhere.

The team base their design search on three simple principles that emerge from the cuckoo’s strategy:

* First, each cuckoo lays one egg (a design solution) at a time, and dumps it in a randomly chosen nest.
* Second, the best nests with a high quality egg (better solution) carry over to the next generation.
* Third, the number of available host nests is fixed, and a host and there is a finite probability of the cuckoo in the nest being discovered.

The team have encapsulated these three principles in a mathematical formula that they then converted to computer software code. The various design parameters and constraints are fed to the software, which tests each “egg” discarding some based on lack of fitness and sending the successful solutions through a second round and so on until an optimal solution emerges.

The team has carried out standard mathematical design tests on their cuckoo search, which itself has now been optimised and also compared it with particle swarm optimisation and other techniques to show that it is more efficient than these other approaches to engineering design of a welded beam and a spring, two key engineering components of many structures. Read more

More information:
Engineering Optimisation by Cuckoo Search ( arXiv paper )
Other papers by Xin-She Yang

Electric Sheep is a collaborative abstract artwork founded by Scott Draves. It’s run by thousands of people all over the world, and can be installed on any ordinary PC or Mac. When these computers “sleep”, the Electric Sheep comes on and the computers communicate with each other by the internet to share the work of creating morphing abstract animations known as “sheep”. The result is a collective “android dream”, an homage to Philip K. Dick’s novel Do Androids Dream of Electric Sheep.

Anyone watching one of these computers may vote for their favorite animations using the keyboard. The more popular sheep live longer and reproduce according to a genetic algorithm with mutation and cross-over. Hence the flock evolves to please its global audience. You can also design your own sheep and submit them to the gene pool.

Electric Sheep

More information:
The Electric Sheep Screen-Saver: A Case Study in Aesthetic Evolution (pdf)
Source code at Source Forge
Electric Sheep Wiki
Artificial Evolution for Computer Graphics, Karl Sims
BodyLab Researching the evolution of the human body shape

I ‘ve been re-reading ‘Out of Control, The Biology of Machines’ and took some time to look up some of the people mentioned who are creating virtual plant life.

If you are interested in creating plants for artificial worlds, or to use to study botany you’ll want to start with the Algorithmic Botany website. There are tons of papers you can download, a book ‘The Algorithmic Beauty of Plants’ and every thing you need to know about ‘cpfg’ the Plant and Fractal generator with Continuous Parameters. Which is a program you can use to generate your own plants using the L-system ( Lindenmayer System ).

If you’d like to dig into the L grammar itself as well as generate plants check out GroIMP, open source Java 3D software for modeling growth grammars.

More information:
Wikipedia, L-system