Monday, December 20, 2004
In evolutionary algorithms a critical parameter that must be tuned is that of selection pressure. If it is set too low then the rate of convergence towards the optimum is likely to be slow. Alternatively if the selection pressure is set too high the system is likely to become stuck in a local optimum due to a loss of diversity in the population. The recent Fitness Uniform Selection Scheme (FUSS) is a conceptually simple but somewhat radical approach to addressing this problem - rather than biasing the selection towards higher fitness, FUSS biases selection towards sparsely populated fitness levels.
A more serious problem is the evolutionary information gathering process.
. . .In short: Intelligence could be something complicated and evolution toward it from an even cleverly designed algorithm of size O(1) could be too slow. As evolution has already taken place, we could add the information from our genes or brain structure to any/our AI system, but this means that the important part is still missing and that it is principally impossible to derive an efficient algorithm from a simple formal definition of AI. Consciousness. For what is probably the biggest question, that of consciousness, we want to give a physical analogy. Quantum (field) theory is the most accurate and universal physical theory ever invented. Although already developed in the 1930s the big question, regarding the interpretation of the wave function collapse, is still open. Although ex-tremely interesting from a philosophical point of view, it is completely irrelevant from a practical point of view. We believe the same to be valid for consciousness in the field of Artificial Intelligence. Philosophically highly interesting but practically unimportant. Whether consciousness will be explained some day is another question.
Links to this post:
Comments: Post a Comment