Showing papers on "Evolutionary programming published in 1986"

Darwinian evolution as a paradigm for AI research

Abstract: Modern Darwinian evolutionary theory is a robust set of concepts that must be transported as a whole if they are to be used paradigmatically in any other context than that of their origin. For Al, the most immediately relevant lessons of evolutionary theory and process are listed below. Their meaning for AI research is that evolution can provide a paradigm for and the outline of an evolving system beyond the metaphorical sense of "evolution," if the full Darwinian paradigm is adopted. This discussion outlines how AI research can make use of the full Darwinian evolutionary paradigm by constructing "learning systems" upon a simulated genetic basis. It also suggests how the Darwinian evolutionary process is already active in one non-biotic context, namely, large organizations (public and private) that extensively use data base management systems (DBMS).

Computational characterization of a hybrid biochip

Abstract: Inherent limitations of conventional silicon-based computing has resulted in much recent research in alternative computing techniques. One such alternative is the "biochip" approach. A review of current work and literature indicates three groups of biocomputing devices. The first group (usually called implantations) is based on utilizing conventional computing techniques in biomedical applications. The second group is geared towards the synthesis of an organic switch similar to the semiconductor PN switch. The third approach (which includes this research) would utilize computing techniques which are characteristic of living organisms. This research investigates an architecture designed to utilize the natural computing techniques of parallelism, fuzziness, pattern recognition, and adaptability. A hybrid model combining conventional silicon techniques and natural techniques is proposed. The model consists of a network of silicon controllers and organic transducing tissues. A computer simulation is then developed to investigate the model. Experiments with the simulation confirm the belief that natural techniques can successfully be used to solve a certain class of problems. Simple but suggestive computing tasks were solved with the model. The programming technique used (evolutionary programming) is treated in detail. The major conclusions of the research are: (1) Conventional silicon computing resources are inefficient at simulating biological processes. Such simulations can however still be advantageously used in solving certain problems. (2) The rigidity and high precision of conventional computing may be relaxed in certain cases. A certain amount of impreciseness can be advantageous in such cases. (3) Silicon and organic computing devices supplement rather than replace each other. (4) Massive parallelism at a low hierarchical level can be used in novel ways which differ from conventional precise operations like matrix multiplications. (5) Simple programming language constructs utilizing natural computing techniques can be implemented as extensions to conventional programming languages. (6) Most of the major obstacles confronting the PN-switch approach to biochip design can be circumvented by the alternative approach of utilizing natural computing techniques. (7) Novel computer architectures can be designed to take advantage of natural computing techniques. (8) Linear topologies may be better than two-dimensional topologies in certain adaptive network applications.

Evolutionary method in discrete programming

Abstract: The paper deals with the two-level partially stochastic optimization method named evolutionary method. The aim of the paper is to present a formal description of the method and its application in discrete programming. The algorithm for planning of investments in chemical industry, based on the method, has been presented.