Artificial K-Lines and Applications - Advances in Computational Science and Engineering - page 85

Artificial K-Lines and Applications
Anestis A. Toptsis and Alexander Dubitski
Dept. of Computer Science and Engineering,
York University, Toronto, Ontario, Canada
{anestis,dubitski}@yorku.ca
Abstract. We propose Artificial K-lines (AKL), a structure that can be used to
capture knowledge through events associated by causality. Like Artificial Neu-
ral Networks (ANN), AKL facilitates learning by capturing knowledge based
on training. Unlike and perhaps complimentary to ANN, AKL can combine
knowledge from different domains and also it does not require that the entire
knowledge base is available prior to the AKL usage. We present AKL and illus-
trate its workings for applications through two examples. The first example
demonstrates that our structure can generate a solution where most other known
technologies are either incapable of, or very complicated in doing so. The sec-
ond example illustrates a novel, human-like, way of machine learning.
1 Introduction
Artificial Intelligence (AI) is enjoying a renewed interest which makes its presence
welcome in many aspects of our daily life. Even before, and certainly since the ap-
pearance of the phrase “AI”, the following questions are of utmost importance: How
come people are able to learn so much ? How come people are creative (i.e., able to
perform a new task, different from two or more previously learned tasks, by being
“inspired” by their previous experiences and knowledge)? These issues have puzzled
philosophers and cognitive scientists for many years, and with the appearance of AI
as a field related to those disciplines, they are among the core AI questions as well.
Numerous attempts to provide an overall answer to these issues, failed during the past
50 years. However, several “theories of memory” have emerged. Notable example are
[1], [2], [3], and [4]. None of these approaches has been fully implemented to date;
however, there have been several reports toward this end. Examples are [5] and [6]. A
central theme in [1], [2], and [3] is the concept of K-lines. Quoting from [1],
When you “get an idea,” or “solve a problem” […] you create what we shall
call a K-line. […]…When that K-line is later “activated”, it reactivates […]
mental agencies, creating a partial mental state “resembling the original”.
In some of our previous works (e.g. [7]) we have used a form of K-lines to address
media handling issues in affective computing systems. In this paper, inspired by the
concept of K-lines, we introduce a structure that can possibly exhibit the caliber of
intelligence usually attributed to Artificial Neural Networks (ANN). To the best of
our knowledge, no such proposal exists for using K-lines in the way described here.
The rest of the paper is organized as follows.