Fabio Caraffini

Parallel memetic structures

Memetic Computing (MC) structures are algorithms composed of heterogeneous operators (memes) for solving optimization problems. In order to address these problems, this study investigates and proposes a simple yet extremely efficient structure, namely Parallel Memetic Structure (PMS). PMS is a single solution optimization algorithm composed of tree operators, the first one being a stochastic global search which explores the entire decision space searching for promising regions. In analogy with electrical networks, downstream of the global search component there is a parallel of two alternative elements, i.e. two local search algorithms with different features in terms of search logic, whose purpose is to refine the search in the regions detected by the upstream element. The first local search explores the space along the axes, while the second performs diagonal movements in the direction of the estimated gradient. The PMS algorithm, despite its simplicity, displays a respectable performance compared to that of popular meta-heuristics and modern optimization algorithms representing the state-of-the-art in the field. Thanks to its simple structure, PMS appears to be a very flexible algorithm for various problem features and dimensionality values. Unlike modern complex algorithm that are specialized for some benchmarks and some dimensionality values, PMS achieves solutions with a high quality in various and diverse contexts, for example both on low dimensional and large scale problems. An application example in the field of magnetic sensors further proves the potentials of the proposed approach. This study confirms the validity of the Ockhams Razor in MC: efficiently designed simple structures can perform as well as (if not better than) complex algorithms composed of many parts.

An analysis on separability for Memetic Computing automatic design

This paper proposes a computational prototype for automatic design of optimization algorithms. The proposed scheme makes an analysis of the problem that estimates the degree of separability of the optimization problem. The separability is estimated by computing the Pearson correlation indices between pairs of variables. These indices are then manipulated to generate a unique index that estimates the separability of the entire problem. The separability analysis is thus used to design the optimization algorithm that addresses the needs of the problem. This prototype makes use of two operators arranged in a Parallel Memetic Structure. The first operator performs moves along the axes while the second simultaneously perturbs all the variables to follow the gradient of the fitness landscape. The resulting algorithmic implementation, namely Separability Prototype for Automatic Memes (SPAM), has been tested on multiple testbeds and various dimensionality levels. The proposed computational prototype proved to be a flexible and intelligent framework capable to learn from a problem and, thanks to this learning, to outperform modern meta-heuristics representing the-state-of-the-art in optimization.

Multi-strategy coevolving aging particle optimization

We propose Multi-Strategy Coevolving Aging Particles (MS-CAP), a novel population-based algorithm for black-box optimization. In a memetic fashion, MS-CAP combines two components with complementary algorithm logics. In the first stage, each particle is perturbed independently along each dimension with a progressively shrinking (decaying) radius, and attracted towards the current best solution with an increasing force. In the second phase, the particles are mutated and recombined according to a multi-strategy approach in the fashion of the ensemble of mutation strategies in Differential Evolution. The proposed algorithm is tested, at different dimensionalities, on two complete black-box optimization benchmarks proposed at the Congress on Evolutionary Computation 2010 and 2013. To demonstrate the applicability of the approach, we also test MS-CAP to train a Feedforward Neural Network modeling the kinematics of an 8-link robot manipulator. The numerical results show that MS-CAP, for the setting considered in this study, tends to outperform the state-of-the-art optimization algorithms on a large set of problems, thus resulting in a robust and versatile optimizer.

Cluster-Based Population Initialization for differential evolution frameworks

This article proposes a procedure to perform an intelligent initialization for population-based algorithms. The proposed pre-processing procedure, namely Cluster-Based Population Initialization (CBPI) consists of three consecutive stages. At the first stage, the individuals belonging to a randomly sampled population undergo two subsequent local search algorithms, i.e. a simple local search that performs moves along the axes and Rosenbrock algorithm. At the second stage, the solutions processed by the two local searches undergo the K-means clustering algorithm and are grouped into sets on the basis of their euclidean distance. At the third stage the best individuals belonging to each cluster are saved into the initial population of a generic optimization algorithm. If the population has not been yet filled, the other individuals of the population are sampled within the clusters by using a fitness-based probabilistic criterion. This three stage procedure implicitly performs an initial screening of the problem features in order to roughly estimate the most interesting regions of the decision space. The proposed CBPI has been tested on multiple classical and modern Differential Evolution variants, on a wide array of test problems and dimensionality values as well as on a real-world problem. The proposed intelligent sampling appears to have a significant impact on the algorithmic functioning as it consistently enhances the performance of the algorithms with which it is integrated.

Super-fit Multicriteria Adaptive Differential Evolution

This paper proposes an algorithm to solve the CEC2013 benchmark. The algorithm, namely Super-fit Multicriteria Adaptive Differential Evolution (SMADE), is a Memetic Computing approach based on the hybridization of two algorithmic schemes according to a super-fit memetic logic. More specifically, the Covariance Matrix Adaptive Evolution Strategy (CMAES), run at the beginning of the optimization process, is used to generate a solution with a high quality. This solution is then injected into the population of a modified Differential Evolution, namely Multicriteria Adaptive Differential Evolution (MADE). The improved solution is super-fit as it supposedly exhibits a performance a way higher than the other population individuals. The super-fit individual then leads the search of the MADE scheme towards the optimum. Unimodal or mildly multimodal problems, even when non-separable and ill-conditioned, tend to be solved during the early stages of the optimization by the CMAES. Highly multi-modal optimization problems are efficiently tackled by SMADE since the MADE algorithm (as well as other Differential Evolution schemes) appears to work very well when the search is led by a super-fit individual.

Multicriteria adaptive differential evolution for global numerical optimization

Differential evolution DE has become a prevalent tool for global optimization problems since it was proposed in 1995. As usual, when applying DE to a specific problem, determining the most proper strategy and its associated parameter values is time-consuming. Moreover, to achieve good performance, DE often requires different strategies combined with different parameter values at different evolution stages. Thus integrating several strategies in one algorithm and determining the application rate of each strategy as well as its associated parameter values online become an ad-hoc research topic. This paper proposes a novel DE algorithm, called multicriteria adaptive DE MADE, for global numerical optimization. In MADE, a multicriteria adaptation scheme is introduced to determine the trial vector generation strategies and the control parameters of each strategy are separately adjusted according to their most recently successful values. In the multicriteria adaptation scheme, the impacts of an operator application are measured in terms of exploitation and exploration capabilities and correspondingly a multi-objective decision procedure is introduced to aggregate the impacts. Thirty-eight scale numerical optimization problems with various characteristics and two real-world problems are applied to test the proposed idea. Results show that MADE is superior or competitive to six well-known DE variants in terms of solution quality and convergence performance.

A CMA-ES super-fit scheme for the re-sampled inheritance search

The super-fit scheme, consisting of injecting an individual with high fitness into the initial population of an algorithm, has shown to be a simple and effective way to enhance the algorithmic performance of the population-based algorithm. Whether the super-fit individual is based on some prior knowledge on the optimization problem or is derived from an initial step of pre-processing, e.g. a local search, this mechanism has been applied successfully in various examples of evolutionary and swarm intelligence algorithms. This paper presents an unconventional application of this super-fit scheme, where the super-fit individual is obtained by means of the Covariance Adaptation Matrix Evolution Strategy (CMA-ES), and fed to a single solution local search which perturbs iteratively each variable. Thus, compared to other super-fit schemes, the roles of super-fit individual generator and global optimizer are switched. To prevent premature convergence, the local search employs a re-sampling mechanism which inherits parts of the best individual while randomly sampling the remaining variables. We refer to such local search as Re-sampled Inheritance Search (RIS). Tested on the CEC 2013 optimization benchmark, the proposed algorithm, named CMA-ES-RIS, displays a respectable performance and a good balance between exploration and exploitation, resulting into a versatile and robust optimization tool.

Re-sampled inheritance search: high performance despite the simplicity

This paper proposes re-sampled inheritance search (RIS), a novel algorithm for solving continuous optimization problems. The proposed method, belonging to the class of Memetic Computing, is very simple and low demanding in terms of memory employment and computational overhead. The RIS algorithm is composed of a stochastic sample mechanism and a deterministic local search. The first operator randomly generates a solution and then recombines it with the best solution detected so far (inheritance) while the second operator searches in an exploitative way within the neighbourhood indicated by the stochastic operator. This extremely simple scheme is shown to display a very good performance on various problems, including hard to solve multi-modal, highly-conditioned, large scale problems. Experimental results show that the proposed RIS is a robust scheme that competitively performs with respect to recent complex algorithms representing the-state-of-the-art in modern continuous optimization. In order to further prove its applicability in real-world cases, RIS has been used to perform the control system tuning for yaw operations on a helicopter robot. Experimental results on this real-world problem confirm the value of the proposed approach.

Compact Differential Evolution Light: High Performance Despite Limited Memory Requirement and Modest Computational Overhead

Compact algorithms are Estimation of Distribution Algorithms which mimic the behavior of population-based algorithms by means of a probabilistic representation of the population of candidate solutions. These algorithms have a similar behaviour with respect to population-based algorithms but require a much smaller memory. This feature is crucially important in some engineering applications, especially in robotics. A high performance compact algorithm is the compact Differential Evolution (cDE) algorithm. This paper proposes a novel implementation of cDE, namely compact Differential Evolution light (cDElight), to address not only the memory saving necessities but also real-time requirements. cDElight employs two novel algorithmic modifications for employing a smaller computational overhead without a performance loss, with respect to cDE. Numerical results, carried out on a broad set of test problems, show that cDElight, despite its minimal hardware requirements, does not deteriorate the performance of cDE and thus is competitive with other memory saving and population-based algorithms. An application in the field of mobile robotics highlights the usability and advantages of the proposed approach.

Memory-saving memetic computing for path-following mobile robots

In this paper, a recently proposed single-solution memetic computing optimization method, namely three stage optimization memetic exploration (3SOME), is used to implement a self-tuning PID controller on board of a mobile robot. More specifically, the optimal PID parameters minimizing a measure of the following error on a path-following operation are found, in real-time, during the execution of the control loop. The proposed approach separates the control and the optimization tasks, and uses simple operating system primitives to share data. The system is able to react to modifications of the trajectory, thus endowing the robot with intelligent learning and self-configuration capabilities. A popular commercial robotic tool, i.e. the Lego Mindstorms robot, has been used for testing and implementing this system. Tests have been performed both in simulations and in a real Lego robot. Experimental results show that, compared to other online optimization techniques and to empiric PID tuning procedures, 3SOME guarantees a robust and efficient control behaviour, thus representing a valid alternative for self-tuning control systems.