The Art of Evolutionary Computation

Computer scientists aren’t particularly considered to be the most artistically oriented. At first glance, it is hard to imagine how the awe striking painting of Mona Lisa could be associated with stacks of impenetrable lines of code vaguely allusive to the English language. Neither is it obvious why computer algorithms and theory of evolution should be in the same sentence. Finally, how could you possibly connect Einstein’s theory of relativity to evolution, art and computer programming? However as a special major in neuroscience and artificial intelligence, I say “Yes, we can!” In my research, I get to use programming techniques inspired by Darwin, chop up masterpieces in art and play around with the pieces combining them in different spatial and temporal dimensions to create art.

The project “Colormosaic” started in a course I took at Hampshire called “Radical Innovation in Digital Arts. 5000 particles (a circle or a rectangle) of various colors would swirl in a random fashion until a key is pressed and they all align themselves to form the specific target image which was imported into the program. During a casual conversation, I heard about a programming technique called genetic algorithm, inspired from Darwin’s ideas about evolution – survival of the fittest and natural selection. The field of evolutionary computation has added many valuable stochastic techniques to our problem-solving toolkit. It has diverse applications, such as in architecture to breed better buildings, in molecular biology to find the lowest energy protein configurations, and in artificial intelligence to teach robots how to walk!

The summer after taking this course I trained myself in GA by reading available literature. I used GA to implement an evolving colormosaic. Initially, I generate a population of say 300 images with random colors assigned to each particle. Their fitness is their resemblance to a given target image. Parents are chosen probabilistically on the basis of their fitness value. Each pair of parents exchange two parts of the image (analogous to crossing over of genes) and the resulting offspring are changed randomly by a tiny bit (mutated). Thus a new population is generated on the principle that the more adaptive organisms have a greater probability of passing on their genes to the subsequent generations. This process is repeated over hundreds of generations, by the end of which we get images that closely resemble the original. Admittedly climatic, this however is not the end of the story.

The following J-term, I took a course in Einstein’s relativity. Inspired by how the boldness and power of the relativistic equation in delineating a limit to nothing less than the speed of light itself, I began finding its parallels in GA. Observe that we have a similar limit – a 100% similarity to the target results in maximum possible fitness. The Lorenz factor, that appears in several equations in special relativity, including time dilation, length contraction, and the relativistic mass formula was easily modified. I incorporated it as a Relative Scaling Factor (RF) in my evolving colormosaic. After all, nature reveals its interconnectedness time and again in mysterious ways.

This brings us to the grand finale of this saga of interdisciplinary research. I ran experiments to on this simple GA application to analyze any advantages of using RF and sure enough. RF scaling enhanced the efficiency and performance of the algorithm. Owing to the generous support from the LITS department, specifically Jason Proctor and Sarah Oelker, my paper was accepted at GECCO (Genetic and Evolutionary Computation conference) where I presented my research this summer in Montreal. I was the only woman at the graduate and undergraduate level. I had an opportunity to meet some of the renowned figures including popular science author and computer scientist Peter Bentley and John Holland who pioneered the field in the 1970s! I was proud to represent MHC. I realized what an incredibly omnipresent network of connections to MHC I have when during a conversation with another attendee, I found out the perhaps unrelated but interesting information that he dated someone from MHC. I had a truly amazing experience at the conference and visiting such a vibrant city as Montreal. The sounds and cheers from the Montreal International Jazz festival would float through my window since my residence happened to be next to some of the concerts. The most vivid memory is of the breathtaking view of fireworks across the lake go off to synchronized music, interspersed with streaks of lightning and roars of thunder. For all the wonderful and intelligent people reading this article, I hope to convey the following heartfelt message through my story and through the following quote by Eleanor Roosevelt: “Life is what you make it. Always has been, always will be. ”


  1. Relative fitness scaling for improving efficiency of proportionate selection in genetic algorithms
  2. Hod Lipson builds “self-aware” robots
  3. Evolution of Mona Lisa

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s