MBL 617E Special Topics in Architectural Design Computing

Assoc. Prof. Dr. Michael Stefan Bittermann

In this course the concept of computational cognition has been elucidated and its relevance in the context of design theory has been studied. The definition of cognition forming the base of the course is: to possess the knowledge of relations among entities in any context and manifesting it by a best action in the same context. Based on a cognition methodology, in which this definition is realized in computational form, a number of well-known characteristics of architectural design process are analysed.

Eyüp Özkan

Image: Excerpt from his analysis of a logical AND operation for three fuzzy premises (0.2, 0.5, 0.9) implemented via a neural computing based on likelihood.

Meryem Yabanigül

Excerpt from her review of a paper on a generic cognitive computing methodology:

Cognition consists of three main topics and it is not possible to mention the cognition and cognitive processes without any of these. …

1. Entities
2. Knowledge
3. Best action

There is no limitation for a group of assets in the definition of the entity. An entity could be a being, object or idea. The cognitive process begins with the interaction of the entities and through the interaction, … entities [are] able to comprehend the incident, context and environment, syllogize with the prior knowledge and generate action. The resulting action is the final and optimal step of the process, therefore the action is the best action for the incident and context.

Sheida Shakeri

Introduction of her explanation of the difference between constraint function and objective function:

Optimization refers to finding the values of inputs in such a way that we get the “best” output values. The definition of “best” varies from problem to problem, but in mathematical terms, it refers to maximizing or minimizing one or more objective functions…. For a single objective optimization problem, the optimum is the minimum or maximum of the objective function. Real-world problems often have several criteria that sometimes conflict… . “

Oğuz Emre Bal

Excerpt from his review of a paper on a generic cognitive computing methodology:

[Cognition] is an action that is chosen from a possibility space that …shows that the decision-maker has information about the whole context and capable to decide what to do. For example, if a dog is taken as a cognitive being, when the dog sees a treat, the dog starts to move to get the reward. This movement of the dog shows that the dog has recognition to the treat and manifests that coming closer, is the best action for the whole context. For another example, for an investor; selling an asset which loses its value is another way to show cognition.

Turan Altıntaş

Correct answers given by the student during the final exam for the three questions:
 
‘What values does a trained cognition network (auto-associative Radial Basis Function network) yield at its outputs when…’
1. … the network is stimulated with a non-dominated solution → It will provide almost exactly the same values as those it was stimulated with,
2. … the network is stimulated with a random solution → It will provide the values of a quite different solution that is nearly non-dominated,
3. …the network is stimulated with a solution that is similar to a non-dominated solution that was used during the network training → It will provide values that are quite similar to the stimulus, however the solution will be nearer to the Pareto front compared to the input