Nature Inspired Computational Creativity

The venture of computational creativity interests me because it brings fields of research together that seem sparsely related. Particle swarm optimisation (PSO) likewise came from the work of a social psychologist and an electrical engineer: James Kennedy and Russell C. Eberhart respectively.

I will discuss my thoughts around applying PSO in computational creativity, first listing three general aims described on Wikipedia:

• To construct a program or computer capable of human-level creativity
• To better understand human creativity and to formulate an algorithmic perspective on creative behaviour in humans
• To design programs that can enhance human creativity without necessarily being creative themselves

^{Source: Computational creativity - Wikipedia}

As an approach to mathematical optimisation, PSO doesn’t produce media typical to artistic expression. Instead it describes an algorithm for producing data. In that sense PSO would appear uncreative and dissimilar to typical human-made art.

However parallels can certainly be drawn between PSO and “human-level creativity”. The originators of PSO, like many artists, uniquely interpreted and took inspiration from nature. Their algorithm is based on the beautiful and spectacular phenomenon of birds flocking:

^{Source: Massive flock of birds - Video}

I think tracing the gradual and ever-changing movement of swarm particles offers some ability to predict the flow of a particular painter’s brush on canvas, where SDS for example would exhibit a more abstract visual style than PSO due to the added element of randomness.

If PSO is capable of the temporal aspect of creativity then visualising it with a more novel approach may get even closer to fulfilling the goals of computational creativity. For example a separate genetic algorithm could be used to generate colour palettes and brush qualities to draw with. I am experimenting with PSO and some of these ideas in my ‘image approximation’ project.

As this is my last weekly blog I have to say the best thing I’ve learnt about natural computing is how open swarm intelligence research is. Talking about its history, community and conferences gave me contextual understanding which helped me appreciate SI’s practical and creative applications more. I was particularly inspired to hear that studies are looking find out whether SDS is Turing complete. It is an ‘out there’ idea but I think it would change my philosophy if virtually everything done by computers could be achieved in a swarm intelligence system.

I would have liked to look more into nature-inspired hardware. It came up at the beginning but seeing further examples of it would’ve been fun (plus maybe a practical lab task like random number generation from sensors/CV).

For Natural Computing (IS53052A)