to do

While I finish writing up the associative learning work done over the summer and before I embark head-on into writing a doctoral thesis (which I hope to start in January – leaving me 7-9 months for writing), I would like to discuss various options of small but publishable projects that I could work on meanwhile. Here are the ideas in mind – roughly in order of current interest:

[1] Ultrastable dynamical neural networks: a discussion of Ashby’s concept of an ultrasable system in the context of today’s evolved dynamical system agents – drawing examples from existing and ideally new work designed purposely to facillitate the comparison.

[2] Balance between chaos and stability in evolved internal mechanisms. What does a task/environment need to be so that agents that do different (relatively random) things get awarded? yet perform a task.

[3] ‘Self-motility’: an experiment that would shed light on what people mean by this in a dynamical systems perspective. The main interest here would be to develop a tool (or set of tools) to measure the ‘reactivity’ or ‘internalisation’ of the actions of an agent. Some systems are purely reactive – rocks. Some systems are purely internal. The interesting systems are those that lay in between and manage to move between the extremes of self-generated-movement and environmentally-generated-movements online.

[4] Sensori-motor learning: evolving an agent whose light sensors move about randomly everytime it is ‘born’. To solve a simple phototaxis task it has to first learn the ‘make-sense’ and only then to do the task.


3 thoughts on “to do

  1. Ashby defines ‘the ultrastable system’ as follows: “Two systems of continuous variables (that we called ‘environment’ and ‘reacting part’) interact, so that a primary feedback (through complex sensory and motor channels) exists between them. Another feedback, working intermittently and at a much slower order of speed, goes from the environment to certain continuous variables which in their turn affect some step-mechanisms change value when and only when these variables pass outside given limits. The step-mechanisms affect the reacting part; by acting as parameters to it they determine how it shall react to the environment”. An instantiation of such an ultrastable system is his famous Homeostat. Ashby also describes throughout his book, ‘Design for a Brain‘, how online adaptation from living organisms are related to the mechanistic theoretical construction of the ultrastable machine. One famous example is that of the kitten reaching with its paw towards the fire and learning that it burns if two close and so on. Does this give you some idea?? I know I will have to eventually explain further and give more precise examples, but perhaps this clarifies the context somewhat.. ? In any case, already in his definition one can begin to see some resemblance with the common dynamical neural networks that we evolve in the situtated, embodied, dynamical systems and evolutionary robotics (SEDER) methodology. My motivation is to make the links more explicit , to point them out with specific examples and to generalise the features as more general principles. Updating perhaps the language in accordance to today’s dynamical system’s literature.

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 )

Google+ photo

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

Connecting to %s