Five select papers

For a more complete list, see my google scholar page

  1. Artificial Neurogenesis: An Introduction and Selective Review
    T. Kowaliw, N. Bredeche, S. Chevallier, and R. Doursat, in "Growing Adaptive Machines", Springer, 2014
    A literature review and introduction to a book: we explore the hypothesis that adaptive growth is a means of producing brain-like machines. The emulation of neural development can incorporate desirable characteristics of natural neural systems into engineered designs.

  2. Bias-Variance Decomposition in Genetic Programming
    T. Kowaliw, R. Doursat, Open Mathematics, De Gruyter, 2016
    An empirical study of linear genetic programming through the lens of the bias-variance decomposition, concluding that (a) the variance between runs is primarily due to initialization rather than the selection of training samples, (b) parameters can be reasonably optimized to obtain gains in efficacy, and (c) larger and more diverse function sets are always preferable, it is preferable to concentrate on including functions which provide information useful to the problem at hand. Finally, we informally argue that an absence of genetic convergence is a potential problem for the creation of ''automated scientists''.

  3. Promoting Creative Design in Interactive Evolutionary Computation.
    T. Kowaliw, A. Dorin, and J. McCormack, IEEE Trans. on Evolutionary Computation, 2012
    We develop a notion of "creativity". We explore the definition via a generative art system, one based on ecosystems. Via a user study, we show validity of our notion of creativity relative to two control groups.

  4. Evolutionary Automated Recognition of an Individual's Artistic Style.
    T. Kowaliw, J. McCormack, and A. Dorin, IEEE CEC 2010
    We train our computer vision system to classify comic images by their originating artist. We then derive a distance function which can tell us, given some arbitrary image from google, how close it is to some given artist's style. Some objective evidence is presented that the function operates intuitively. Also: we release a new open database of artistic images.

  5. Environment as a Spatial Constraint on the Growth of Structural Form.
    T. Kowaliw, P. Grogono, and N. Kharma, GECCO 2007
    Exploration of an artificial development for the growth of structural design. The environment in which the designs were grown was perturbed, post-optimization: either be changing the amount of resources, or by changing the geometric shape. The growth procedure can adapt to changes in size, and some genomes can adapt to nearly any environmental geometry. These are artificial analogues to phenotypic robustness and phenotypic plasticity (polymorphism).