NEO 2015

For all supervised learning problems, where the quality of solutions is measured by a distance between target and output values (error), geometric semantic operators of genetic programming induce an error surface characterized by the absence of locally suboptimal solutions (unimodal error surface). So, genetic programming that uses geometric semantic operators, called geometric semantic genetic programming, has a potential advantage in terms of evolvability compared to many existing computational methods. This fosters geometric semantic genetic programming as a possible new state-of-the-art machine learning methodology. Nevertheless, research in geometric semantic genetic programming is still much in demand. This chapter is oriented to researchers and students that are not familiar with geometric semantic genetic programming, and are willing to contribute to this exciting and promising field. The main objective of this chapter is explaining why the error surface induced by geometric semantic operators is unimodal, and why this fact is important. Furthermore, the chapter stimulates the reader by showing some promising applicative results that have been obtained so far. The reader will also discover that some properties of geometric semantic operators may help limiting overfitting, bestowing on genetic programming a very interesting generalization ability. Finally, the chapter suggests further reading and discusses open issues of geometric semantic genetic programming.

Sentiment analysis is one of the most important tasks in text mining. This field has a high impact for government and private companies to support major decision-making policies. Even though Genetic Programming (GP) has been widely used to solve real world problems, GP is seldom used to tackle this trendy problem. This contribution starts rectifying this research gap by proposing a novel GP system, namely, Root Genetic Programming, and extending our previous genetic operators based on projections on the phenotype space. The results show that these systems are able to tackle this problem being competitive with other state-of-the-art classifiers, and, also, give insight to approach large scale problems represented on high dimensional spaces.

This paper presents a genetic programming (GP) approach enhanced with a local search heuristic (GP-LS) to emulate the Doherty 7 W @ 2.11 GHz Radio Frequency (RF) Power Amplifier (PA) conversion curves. GP has been shown to be a powerful modeling tool, but can be compromised by slow convergence and computational cost. The proposal is to combine the explorative search of standard GP, which build the syntax of the solution, with numerical methods that perform an exploitative and greedy local optimization of the evolved structures. The results are compared with traditional modeling techniques, particularly the memory polynomial model (MPM). The main contribution of the paper is the design, comparison and hardware emulation of GP-LS for FPGA real applications. The experimental results show that GP-LS can outperform standard MPM, and suggest a promising new direction of future work on digital pre-distortion (DPD) that requires complex behavioral models.

In this work we propose the implementation of an automatic random tree generator on an FPGA for genetic programming (GP). While most authors in specialized literature avoid the use of the tree data structure in their implementations of GP on Field Programmable Gate Arrays (FPGAs), due to the impossibility of using pointers (references) in the Very High Speed Integrated Circuit Hardware Description Language (VHDL), we propose two methods for a single matrix implementation and one for a vector implementation. All trees in the population are created in concurrent processes leading to significant time savings. We present pseudocode and results of hardware consumption for matrix and vector implementations. Results show that up to 100 trees can be implemented in a Spartan-6 FPGA using the representation of one tree in a single matrix in parallel processes. Moreover, this implementation requires less resources than the apparently simpler vector representation.

The mixed postman problem consists of finding a minimum cost tour of a mixed graph traversing all its vertices, edges, and arcs at least once. We consider an NP-hard variant of the mixed postman problem where all arcs must be traversed exactly once, known as the edges postman problem. We present four approximation algorithms for the edges postman problem with different guarantees. Our main result is a \(\frac{4}{3}\)-approximation algorithm for the edges postman problem.

Premature convergence is one of the most important recurrent drawbacks of Evolutionary Algorithms and other metaheuristics. As a result, several methods to alleviate this problem have been devised. One alternative is to explicitly control the diversity of the population. In this chapter, a recently proposed survivor selection strategy is incorporated into a memetic algorithm and analyzed using three different combinatorial optimization problems. This strategy is based on adopting multi-objective concepts for solving single-objective problems by considering the contribution to diversity as an explicit objective. Additionally, it incorporates the principle of adapting the balance between exploration and exploitation to the different stages of the optimization by taking into account the stopping criterion and elapsed time. These new methods provide important benefits when compared to more mature methods that rely on different principles to delay convergence of the population. Additionally, new best-known solutions are generated for several instances of the problems, thus providing proofs of the considerable benefits and robustness yielded by the schemes that incorporate this novel replacement strategy.

Pharmaceutical researchers and biotechnology companies are devoted to developing medicines, such as: therapeutic proteins, human insulin, vaccines for hepatitis, food grade protein, chymosin detergent enzyme, and cryophilic protease. This allows patients to live longer, healthier, and more productive. Within this context, there is a high degree of consensus in the biomanufacturing industry that product quality, customer service, and cost efficiency are fundamental for success. Based on our knowledge there has not been an adequate flexibility strategy to manufacture different multiproduct drug substances, such as designing a plant, determining the number of units for a specific task, assigning raw materials to different production processes, and deciding the production planning. The aim of this work is to minimize the investment cost and find out the number and size of parallel equipment units in each stage of multiproduct batch plant design (MBPD). For this purpose, it is proposed to solve the problem in two different ways: the first way is by using a particle swarm algorithm (PSA) and the second way is by a genetic algorithm (GA). This paper presents the effectiveness and performance comparison of PSA and GA for optimal design of a MBPD. The experimental results (given by investment cost, number and size of equipment, computational time, and idle times within the plant) obtained by the GA are better than those found by the PSA. This methodology can help the decision makers, and constitutes a very promising framework for finding a set of good solutions.

Multi-objective optimization problems are commonly encountered in real world applications. In some applications, where the gradient information of the objective functions is available, it is natural to consider a gradient-based multi-objective optimization algorithm for relatively high convergence speed and stability. In this chapter, we consider a recently proposed gradient-based approach, called the hypervolume indicator gradient ascent method. It is designed to maximize the hypervolume indicator in the steepest direction by calculating its gradient field with respect to decision vectors. The hypervolume indicator gradient derivation will be covered in this chapter. Despite the elegance of this approach, a critical issue arises when applying the gradient computation for some of the decision vectors: the gradient at a dominated point is either zero or undefined, which restricts the usage of this approach. To remedy this, five methods are proposed to provide a search direction for dominated points (at which the hypervolume indicator gradient fails to do so). These five methods are devised for the bi-objective optimization case and are illustrated in detail. In addition, a thorough empirical study is carried out to investigate the convergence behavior of these five methods. The combination of the hypervolume indicator gradient and the proposed five methods constitute a novel gradient-based, bi-objective optimization algorithm, which is validated and benchmarked. The benchmark results show interesting performance comparisons among the five proposed methods.

The most important part of the control design for nonlinear dynamical systems is to guarantee the stability. Then, the control is quantitatively designed to meet multiple and often conflicting performance objectives. The performance of the closed-loop system is a function of various system and control parameters. The quantitative design using multiple parameters to meet multiple conflicting performance objectives is a challenging task. In this chapter, we present the recent results of Pareto optimal design of controls for nonlinear dynamical systems by using the advanced algorithms of multi-objective optimization. The controls can be of linear PID type or nonlinear feedback such as sliding mode. The advanced algorithms of multi-objective optimization consist of parallel cell mapping methods with sub-division techniques. Interesting examples of linear and nonlinear controls are presented with extensive numerical simulations.

This chapter presents an overview of Multi Agent Collaborative Search (MACS), for multi-objective optimisation with an analysis of different heuristics for local search. In particular the effects of simple inertia and differential evolution operators in combination with pattern search and gradient methods are investigated. Different benchmarks are used to demonstrate the effectiveness of the MACS framework and of the heuristics for both local and global search. The MACS framework is tested on two sets of academic problems and two real space mission design problems using the IGD and the success rate as performance metrics. The performance of MACS is compared against three known multi-objective optimisation algorithms: NSGA-II, MOAED and MTS.

This chapter is about Generalized Differential Evolution (GDE), which is a general purpose optimizer for global nonlinear optimization. It is based on Differential Evolution (DE), which has been gaining popularity because of its simplicity and good observed performance. GDE extends DE for problems with several objectives and constraints. The chapter concentrates on describing different development phases and performance of GDE but it also contains a brief listing of other multi-objective DE approaches. Ability to solve multi-objective problems is mainly discussed, but constraint handling and the effect of control parameters are also covered. It is found that the latest GDE version is effective and efficient for solving constrained multi-objective problems having different types of decision variables.

In this chapter we present the adaptions of the recently proposed Directed Search method to the context of unconstrained parameter dependent multi-objective optimization problems (PMOPs). The new method, called \(\lambda \)-DS, is capable of performing a movement both toward and along the solution set of a given differentiable PMOP. We first discuss the basic variants of the method that use gradient information and describe subsequently modifications that allow for a gradient free realization. Finally, we show that \(\lambda \)-DS can be used to understand the behavior of stochastic local search within PMOPs to a certain extent which might be interesting for the development of future local search engines, or evolutionary strategies, for the treatment of such problems. We underline all our statements with several numerical results indicating the strength of the novel approach.

In the year 1929 a German psychiatrist, named Hans Berger, demonstrated for the first time that the electric activity of the human brain was related to the person’s mental state. He also announced the possibility of registering such type of electric activities without opening the human head, i.e. non invasive procedure, and that such electric activities could be plotted on a graph. Berger called such type of registration as electroencephalogram (EEG). EEG signals research has been growing over the years due to the their increasing use to control electronic devices in all sorts of contexts. The present work developed a prototype to control a differential robot by means of EEG signals using the detection of movement intention of the right and left hand. The study covered on one hand, the analysis and design of the teleoperation system, and on the other hand, the robot tele operational tests. It is important to point out that the robot was designed and built to meet the technical research purposes. The programming of the EEG signal processing was made using the API provided by MATLAB. In turn, the programming for controlling the mobile differential robot was made with Wiring and Python. Lastly, several tests and experiments were carried out, and they showed that the objective in view was met.

This chapter proposes the use of a scalable platform to run a complex recommendation system. We focus on a system made up of several recommendation algorithms which are run as an offline process. This offline process generates user profiles that represent which algorithm should provide the recommendations to a given user and item, and will be combined with a fuzzy decision system to generate every recommendation. Yet, given the amount of data that will be processed and the need to run that offline process frequently, we propose to reduce execution time by using Hadoop, a scalable, distributed and fault-tolerant platform. Obtained results shows how the main goal pursued here is achieved: the efficient use of computer resources which allows for a significant reduction in computing time.

In this work we are sizing the transistors of an CMOS Miller amplifier. More precisely, we are addressing the bi-objective problem of maximizing the DC gain and minimizing the current supply of Miller’s amplifier. As we consider a bi-objective problem we get a set of solutions, instead of a single one. Next, we study three schemes to select one or several of those solutions, based on the ones with greater tolerances to variations in the circuit elements, or with lowest sensitivities, or with lowest statistical variations using Monte Carlo method. Also, it is possible to include this tolerance analysis in the optimization loop by adding an objective to the given optimization problem.

This study presents a hybrid of artificial neural network and NSGA-II for multi-objective optimization of a particular plastic injection molding process. The objectives to be optimized are: a dimension of the finished plastic product (product quality), the processing time (productivity), and the energy consumption (manufacturing cost). The data collection and results validation are made on a 330 ton plastic injection machine. The design variables considered are mold temperature, material temperature, injection time, packing pressure, packing pressure time, and cooling time. An artificial neural network is used to map the relationship between design variables and output variables. Then, NSGA-II is used to find the set of Pareto optimal solutions. The results show that the methodology gives the designer flexibility and robustness to choose different scenarios according to current design requirements in terms of quality, productivity and energy savings.

This work studies the ambulance location problem for the Red Cross in Tijuana, Baja California, Mexico. The solution to the ambulance location problem is to optimally locate all available ambulances within the city such that coverage of the city population is maximized and a quick response to any emergency is ensured. The problem is posed using three different coverage models; namely the Location Set Covering Model (LSCM), the Maximal Covering Location Problem (MCLP) and the Double Standard Model (DSM), also we proposed robust versions of each model, where the goal was to find a single solution that might provide optima coverage in several different scenarios. Using real-world data collected from over 44,000 emergency calls received by the Red Cross of Tijuana, several scenarios were generated that provide different perspectives of the demand throughout the city, considering such factors as the time of day, work and off-days, geographical organization and call priority. These scenarios are solved using Integer Linear Programming and the solutions are compared with the current locations used by the Red Cross. Results show that demand coverage and response times can be substantially improved without additional resources.