N. Stepp
Academic Pages
Research Interests
Artificial Intelligence

At a very general level, my interests revolve around how it is that intelligence may emanate from non-intelligence. When this happens in a man-made system, of course, this becomes the study of Artificial Intelligence. That is, however, not the only field asking this question. How one gets intelligence from non-intelligence is a general concern for any field dealing with systems exhibiting intelligent behavior. This sometimes includes the study of Humans.

Being so general, this question is not actually that easy to answer. In fact, it is very hard. My more specific interests attempt to get at small pieces of this very hard problem.

Anticipatory Systems

Anything that acts, if it hopes to act with some purpose, must anticipate the effects of its actions. If I wait until after I reach for my fork to see if I will knock over the wine, my cleaning bill will be very high.

The ability to have some access to the future, then, appears to be an integral part of being a perceiving—acting system. One way to have this access is the use of forward models, along with their attendant problems.

How can we have access to the future without a model? This question leads one to consider so-called Strong Anticipation (Dubois, 2001, 2003).

Graph Dynamics of Perceptual Systems

Organisms are so organized as to take advantage of properties of the world. For a simple example, think phototaxis.

The process of this organization occurs at many time scales (evolution, development, learning, ...). Understanding the principles governing this process at all scales is key to understanding the driving force for life.

Being a bit more down to Earth, the study of changing organization will help to build more intelligent systems. It seems obvious that we are not smart enough to hand tune one outright, but perhaps if we know enough, we might be able to grow one.

Turning Everything into a Vector Space Linear algebra is very general. If you are describing something to me, I'm probably thinking about vector spaces.