Expansion of Neuro-Modules by Structure Evolution
نویسندگان
چکیده
Two methods for the extension of neuro-modules are introduced resulting in a new behavioral functionality. We call them restricted and semi-restricted module expansion. These methods are developed and applied using a modular neuro-dynamics approach to behavior control of autonomous mobile robots. Evolved neuro-controllers which solve an obstacle avoidance task are expanded to show in addition a positive phototropism. All resulting neuro-modules produce a robust light seeking behavior. These neuro-modules use recurrent connectivity structures and non-trivial dynamical features to enable the robot to solve its task. For each neuro-module the structure-function-relation is analyzed. The presented results demonstrate that restricted and semi-restricted expansion are promising methods for generating efficient extensions of recurrent neural networks with additional behavioral functionality.
منابع مشابه
Advanced Dynamic Simulation of Membrane Desalination Modules Accounting for Organic Fouling
A reliable dynamic simulator (based on a sound process model) is highly desirable for optimizing the performance of individual membrane modules and of entire desalination plants. This paper reports on progress toward development of a comprehensive model of the complicated physical-chemical processes occurring in spiral wound membrane (SWM) modules, that accounts for the...
متن کاملNMODE - Neuro-MODule Evolution
Modularisation, repetition, and symmetry are structural features shared by almost all biological neural networks. These features are very unlikely to be found by the means of structural evolution of artificial neural networks. This paper introduces NMODE, which is specifically designed to operate on neuro-modules. NMODE addresses a second problem in the context of evolutionary robotics, which i...
متن کاملEvolving Neuro - Modules and their Interfacesto Control
An evolutionary algorithm for the creation of recurrent network structures is presented. The aim is to develop neural networks controlling the behaviour of miniature robots. Two neuro-modules are created separately using this evolutionary approach. The rst neuro-module gives the agents the ability to move within an environment without colliding with obstacles. The second neuro-module provides t...
متن کاملG-frames in Hilbert Modules Over Pro-C*-algebras
G-frames are natural generalizations of frames which provide more choices on analyzing functions from frame expansion coefficients. First, they were defined in Hilbert spaces and then generalized on C*-Hilbert modules. In this paper, we first generalize the concept of g-frames to Hilbert modules over pro-C*-algebras. Then, we introduce the g-frame operators in such spaces and show that they sha...
متن کاملSearch Space Restriction of Neuro-evolution through Constrained Modularization of Neural Networks
Evolving recurrent neural networks for behavior control of robots equipped with larger sets of sensors and actuators is difficult due to the large search spaces that come with the larger number of input and output neurons. We propose constrained modularization as a novel technique to reduce the search space for such evolutions. Appropriate neural networks are divided manually into logically and...
متن کامل