Gestalt Aspects of Security Patterns
Mario Köpen
KYUSHU INSTITUTE OF TECHNOLOGY, Japan  

 [Abstract] Gestalt laws of vision are an apparent phenomenon of visual perception that still lack general understanding, despite more than 100 years after its first mentioning in psychological literature have passed. In this contribution, we want to promote Gestalt as challenge to computer science. We want to focus on the observation that Gestalt laws strongly affect the way, in which we organize and structure our living environment, and thus adapting to the needs, possibilities, but also limitations of human cognition. This also refers to forensic case studies, and the ambiguous role of Gestalt there: Gestalt can help to discover and gain evidence from visual examination, but it can also simply cause the overlooking of essential visual cues. An experimental study on information hiding by the method of modifying a logo image will demonstrate that attention for changes is not a linear function of the amount of change. In this sense, a number of (probably even simple) forensic techniques and "rules of thumb" can be also seen as employing Gestalt laws for revealing forensic evidence.

This gives some understanding for an interest in the computational handling of Gestalt. In a second part, the state of research on Gestalt in engineering sciences, esp. image processing and pattern analysis, will be critically reviewed, and their strong and weak points will be evaluated. But moreover, new emerging computational paradigms and models will be evaluated according to what they might provide for the understanding of Gestalt. Among these paradigms and models, we can find the Neural Darwinism, which relates evolutionary concepts to the processing of the brain, the recently proposed Cogency Confabulation, which relates learning with the maximization of a priori probability, and Jeff Hawkins hierarchical temporal memory model of the brain that comes most close to a brain processing model including Gestalt laws "for free," while maintaining a computational model at the same time.

 [Biography]

Ensemble Methods For Evolutionary Algorithms
Ponnuthurai Nagaratnam Suganthan
NTU, Singapore  

 [Abstract]Over the last 4-5 decades, evolutionary computation researchers have proposed several alternative approaches to construct evolutionary algorithms (EAs). Some such alternatives are one-point / two-points / uniform crossover operators, tournament / ranking / stochastic uniform sampling selection methods, clearing / crowding / sharing based niching algorithms, adaptive penalty / epsilon / superiority of feasible constraint handling approaches and so on. Clearly, there are several alternative approaches at every step of an EA and users will have to perform numerous simulations to pick the best approaches. In addition, each approach may require users to fine tune associated parameters. Furthermore, at different stages of evolution, different operators and different parameter values may be more appropriate. Therefore, the trial and error approach to operator selection and associated parameter tuning is not efficient. Recently, we proposed an ensemble strategy to benefit from both the availability of diverse operators and wide range of the associated parameters. Our research has shown the general applicability of the ensemble strategy in solving diverse problems by using different populated optimization algorithms. This talk will present some of our resent results along this novel research direction.

 [Biography]