Although techniques for using formal speci cations have been progressing, methods for developing formal speci cations themselves have improved little. To alleviate this problem, we propose a parallel re nement approach to speci cation acquisition where the designer uses an object-oriented speci cation representation while an underlying speci cation composition system manipulates an equivalent t...

A method for calculating genetic drift in terms of changing population fitness variance is presented. The method allows for an easy comparison of different selection schemes and exact analytical results are derived for traditional generational selection, steady-state selection with varying generation gap, a simple model of Eshelman’s CHC algorithm, and (μ + λ) evolution strategies. The effects ...

An oscillatory correlation model of human motion perception is proposed based on the integration of motion and luminance information. The model is composed of two parallel pathways that segment the input scene based on motion and luminance, respectively. Combining these segmentations, the model re nes the motion estimates in the integration stage to obtain the nal segmentation in the motion pat...

A new evolution scheme is presented, memorizing the extreme (best and worst) past individuals through distributions over the binary search space. These distributions are used to bias the mutation operator in a (+) Evolution Strategy, guiding the generation of newborn oospring: diierent mimetic strategies are deened, combining either attraction, indiierence or repulsion with respect to the two d...

An evolutionary algorithm based on Evolution Strategy (ES) is presented, which includes time-related command execution and the generation of process control elements. Its concept is enlarged by problem-oriented type definitions for parameters, this has allowed a flexible implementation for different applications. The GLEAM algorithm includes new features which distinguish it from ES and GAs, am...

Deep reinforcement learning (RL) methods generally engage in exploratory behavior through noise injection in the action space. An alternative is to add noise directly to the agent’s parameters, which can lead to more consistent exploration and a richer set of behaviors. Methods such as evolutionary strategies use parameter perturbations, but discard all temporal structure in the process and req...

Fundamental to research at the Institute of Design is generation of a carefully constructed, robust question. This question emerges from the faculty developed research agenda with its five domains of interest matrixed against three research foci. Central to the development of particular research from this matrix is the generation of research questions. The question-based approach provides two s...

Population based methods handle a population of individuals that evolves with the help of information exchange procedures. It should be noted that many different algorithms could be described within this framework. The best known population-based algorithms include genetic algorithms [8], [6] evolution strategies and evolution programming [12], scatter search [3], adaptive memory algorithms [4]...

One feature of evolving populations is that genetic operators act on the genotypic level while selection acts on the phenotypic level. We call the transformation between genes and phenes a developmental process because of its dynamics. Generally, no inferences can be made from phenes to genes, i.e. the mapping from genes to phenes is not an isomorphic one as it is assumed in Genetic Algorithms ...

This memo de nes an architectural model for active networks. It de nes the assumptions and goals for the active network and describes the functionality of each node. The approach speci es a set of basic functions from which the end-to-end abstractions presented to the user are implemented.