Fitness functions for the unconstrained evolution of digital circuits

Evolvable Systems: From Biology to Hardware

Greensted

et al.

Self Cite

Abstract. Evolution is particularly good at finding specific solutions, which are only valid for exactly the input and environment that are presented during evolution. In most evolution experiments the input pattern order problem is not considered, even though the ability to provide a correct result for any input pattern is a prerequisite for valid circuits. Therefore, the importance of including randomness in the input pattern applied during evolution is addressed in this paper. This is shown to be mandatory-particularly in the case of unconstrained intrinsic evolution of digital circuits-in order to find valid solutions. The different ways in which unconstrained evolution and constrained evolution exploit resources of a hardware substrate are compared. It is also shown that evolution benefits from versatile input configurations. Furthermore, hierarchical fitness functions, previously introduced to improve the evolution of combinatorial circuits, are applied to the evolution of sequential circuits.

Section: Fitness Measuring Methodsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

The Input Pattern Order Problem: Evolution of Combinatorial and Sequential Circuits in Hardware

Evolvable Systems: From Biology to Hardware

Greensted

et al.

Self Cite

“…It is the evaluation process that provides the foundation for the selection process, and thereby greatly contributes to guiding evolutionary search through the solution space in order to construct the desired circuits. It was shown earlier that a fitness function, which is more suited for the particular problem at hand had a considerably positive impact on the performance of evolution in designing the desired circuit [11], [18]. The importance of different selection schemes, being the second major driving force in evolutionary algorithms, have also been addressed by various researchers [2], [22].…”

Section: Challenges In the Evaluation Process In Ehwmentioning

confidence: 99%

“…The most likely reasons for getting stuck in local optima during the course of evolutionary search are: first, unsuitable fitness evaluation methods that are not capable of considering special cases within certain problem domains [11]. Second, intermittent solutions that can mislead the search process and cause the EA to stall.…”

Section: A Getting Stuck In Local Optimamentioning

confidence: 99%

The input pattern order problem II: Evolution of multiple-output circuits in hardware

2009 IEEE Workshop on Evolvable and Adaptive Hardware

Miller

et al. 2009

Abstract-It has been shown in previous work that the design of the input pattern plays and important role in intrinsic evolvable hardware. The results have shown that randomising the input pattern during the course of evolution is mandatory in order to achieve valid circuits that work correctly under all conditions. Furthermore, it was found that randomising the input pattern order helps evolution to avoid local optima. Due to the fact that it becomes exponentially more difficult to successfully evolve circuits with increasing number of outputs, this paper addresses the importance of the input pattern order problem (IPOP) for the second time, with a focus on the intrinsic evolution of digital circuits with multiple outputs. In this paper, we investigate whether it is also possible to design the input pattern in a way to improve the performance of evolution when tackling multiple output problems. A different approach to IPOP is taken, where input pattern order is used as a method of balancing the distribution of the output values. The aim is to equalise the fitness value range by using weighted input vectors in order to obtain a smoother search landscape for multiple output circuits. The proposed method is tested on the intrinsic evolution of 4-bit AND, 4-bit parity, 2-bit full adder, and 2-bit multiplier. Furthermore, the results for the intrinsically evolved multipliers are compared with software simulations in order to determine whether there is a benefit for both intrinsic and extrinsic evolution.

“…The organism is allowed to age for 10 developmental steps and then evaluated once at age 10. Hierarchichal Bitstring Sampling (HBS) fitness function is used to check if the developed circuit functions correctly [15]. Therefore, the developmental system goes offline once it reaches maturity in these experiments.…”

Section: Development Of An Even N-bit Parity Circuitmentioning

confidence: 99%

On the properties of artificial development and its use in evolvable hardware

2009 IEEE Symposium on Artificial Life

Miller

et al. 2009

Self Cite

Abstract-The design of a new biologically inspired artificial developmental system is described in this paper. In general, developmental systems converge slower and are more computationally expensive than direct evolution. However, the performance trends of development indicate that the full benefit of development will arise with larger and more complex problems that exhibit some sort of regularity in their structure: thus, the aim is to evolve larger electronic systems through the modularity allowed by development. The hope is that the proposed artificial developmental system will exhibit adaptivity and fault tolerance in the future. The cell signalling and the system of Gene Regulatory Networks present in biological organisms are modelled in our developmental system, and tailored for tackling real world problems on electronic hardware. For the first time, a Gene Regulatory Network system is successfully shown to develop the complete circuit structure of a desired digital circuit without the help of another mechanism or any problem specific structuring. Experiments are presented that show the modular behaviour of the developmental system, as well as its ability to solve non-modular circuit problems.