Abstract This paper explores a novel technique for learning the fitness function for search algorithms such as evolutionary strategies and hillclimbing. The aim of the new technique is to learn a fitness function (called a Learned Guidance Function) from a set of sample solutions to the problem. These functions are learned using a supervised learning approach based on deep neural network learning, that is, neural networks with a number of hidden layers. This is applied to a test problem: unscrambling the Rubik's Cube using evolutionary and hillclimbing algorithms. Comparisons are made with a previous LGF approach based on random forests, with a baseline approach based on traditional error‐based fitness, and with other approaches in the literature. This demonstrates how a fitness function can be learned from existing solutions, rather than being provided by the user, increasing the autonomy of AI search processes.

In the path planning task for autonomous mobile robots, robots should be able to plan their trajectory to leave the start position and reach the goal, safely. There are several path planning approaches for mobile robots in the literature. Ant Colony Optimization algorithms have been investigated for this problem, giving promising results. In this paper, we propose the Max-Min Ant System for Dynamic Path Planning algorithm for the exploratory path planning task for autonomous mobile robots based on topological maps. A topological map is an environment representation whose focus is the main reference points of the environment and their connections. Based on this representation, the path can be composed by a sequence of state/actions pairs, which facilitates the navigability of the path, with no need to have the information of the complete map. The proposed algorithm was evaluated in static and dynamic environments, showing promising results in both of them. Experiments in dynamic environments show the adaptability of our proposal.

In this paper, we discuss students' expectations and experiences in the first term of the Year in Computing, a new programme for non-computing majors at the University of Kent, a public research university in the UK. We focus on the effect of students' home discipline on their experiences in the programme and situate this work within the context of wider efforts to make the study of computing accessible to a broader range of students.

This work proposes a novel Situation-Aware FEar Learning (SAFEL) model for robots. SAFEL combines concepts of situation-aware expert systems with well-known neuroscientific findings on the brain fear-learning mechanism to allow companion robots to predict undesirable or threatening situations based on past experiences. One of the main objectives is to allow robots to learn complex temporal patterns of sensed environmental stimuli and create a representation of these patterns. This memory can be later associated with a negative or positive “emotion”, analogous to fear and confidence. Experiments with a real robot demonstrated SAFEL's success in generating contextual fear conditioning behavior with predictive capabilities based on situational information.

In the path planning problem for autonomous mobile robots, robots have to plan their path from the start position to the goal. In this paper, we investigate the application of the MMAS algorithm to the exploratory path planning problem, in which the robots should explore the environment at the same time they plan the path. Max-min ant system is an ant colony optimization algorithm that exploits the best solutions found. In addition, to analyze the quality of solutions obtained, we also analyze the traveled distance spent by robots in the first iteration of the algorithm. The environment is previously unknown to the robots, although it is represented by a topological map, that does not require precise information from the environment and provides a simple way to execute the navigation of the path. Thus, the paths are represented by a sequence of actions that the robots should execute to reach the goal. The navigation of the best solution found was implemented in a realistic robotic simulator. The proposed algorithm provides a very good performance in relation to a genetic algorithm and the well-known A∗ algorithm that deal with this problem.

In this paper, we optimize the predictive performance of a Situation-Aware FEar Learning model (SAFEL) by investigating the relationship between its parameters. SAFEL is a hybrid computational model based on the fear-learning system of the brain, which was developed to provide robots with the capability to predict threatening or undesirable situations based on temporal context. The main aim of this work is to improve SAFEL's emotional response. An emotional response coherent with environmental changes is essential not only for self-preservation and adaptation purposes, but also for improving the believability and interaction skills of companion robots. Experiments with a NAO humanoid robot show that adjusting the ratio between two parameters of SAFEL can significantly increase the predictive performance and reduce parameter settings.

This work proposes a theoretical architectural model based on the brain's fear learning system with the purpose of generating artificial fear conditioning at both stimuli and context abstraction levels in robot companions. The proposed architecture is inspired by the different brain regions involved in fear learning, here divided into four modules that work in an integrated and parallel manner: the sensory system, the amygdala system, the hippocampal system and the working memory. Each of these modules is based on a different approach and performs a different task in the process of learning and memorizing environmental cues to predict the occurrence of unpleasant situations. The main contribution of the model proposed here is the integration of fear learning and context awareness in order to fuse emotional and contextual artificial memories. The purpose is to provide robots with more believable social responses, leading to more natural interactions between humans and robots.

The prediction of protein secondary structure is a topic that has been tackled by many researchers in the field of bioinformatics. In previous work, this problem has been solved by various methods including the use of traditional classification neural networks with the standard error back-propagation training algorithm. Since the traditional neural network may have a poor generalisation, the Bayesian technique has been used to improve the generalisation and the robustness of these networks. This paper describes the use of optimised classification Bayesian neural networks for the prediction of protein secondary structure. The well-known RS126 dataset was used for network training and testing. The experimental results show that the optimised classification Bayesian neural network can reach an accuracy greater than 75%.

Ant colony optimization (ACO) algorithms have been successfully applied to discover a list of classification rules. In general, these algorithms follow a sequential covering strategy, where a single rule is discovered at each iteration of the algorithm in order to build a list of rules. The sequential covering strategy has the drawback of not coping with the problem of rule interaction, i.e., the outcome of a rule affects the rules that can be discovered subsequently since the search space is modified due to the removal of examples covered by previous rules. This paper proposes a new sequential covering strategy for ACO classification algorithms to mitigate the problem of rule interaction, where the order of the rules is implicitly encoded as pheromone values and the search is guided by the quality of a candidate list of rules. Our experiments using 18 publicly available data sets show that the predictive accuracy obtained by a new ACO classification algorithm implementing the proposed sequential covering strategy is statistically significantly higher than the predictive accuracy of state-of-the-art rule induction classification algorithms.

Variables are a fundamental component of computer programs. However, rarely has the construction of new variables been left to the evolutionary process of a tree-based Genetic Programming system. We present a series of modifications to an existing GP approach to allow the evolution of high-level imperative programs with limited scope variables. We make use of several new program constructs made possible by the modifications and experimentally compare their use. Our results suggest the impact of variable declarations is problem dependent, but can potentially improve performance. It is proposed that the use of variable declarations can reduce the degree of insight required into potential solutions.

This paper explores the idea of applying evolutionary algorithms to those search spaces that are defined extensionally, i.e. by listing every item in the space. When these spaces are with a function that returns similar elements given a key element, analogies of mutation and crossover can be defined. This idea is discussed in general, and specific examples are given where the search is for images, in particular where image search is carried out using an interactive genetic algorithm.

Genetic programming has proven capable of evolving solutions to a wide variety of problems. However, the successes have largely been with programs without iteration or recursion; evolving recursive programs has turned out to be particularly challenging. The main obstacle to evolving recursive programs seems to be that they are particularly fragile to the application of search operators: a small change in a correct recursive program generally produces a completely wrong program. In this paper, we present a simple and general method that allows us to pass back and forth from a recursive program to an associated non-recursive program. Finding a recursive program can be reduced to evolving non-recursive programs followed by converting the optimum non-recursive program found to the associated optimum recursive program. This avoids the fragility problem above, as evolution does not search the space of recursive programs. We present promising experimental results on a test-bed of recursive problems.

Cartesian Genetic Programming (CGP) is a form of Genetic Programming that uses directed graphs to represent programs. In this paper we propose a way of structuring a CGP algorithm to make use of the multiple phenotypes which are implicitly encoded in a genome string. We show that this leads to a large increase in efficiency compared with standard CGP where genomes are translated into only one phenotype. We call this method Multi Expression CGP (ME-CGP), based on Mihai Oltean's work on Multi Expression Programming using linear GP.

This paper investigates the effect of different levels of rationality on the Lucas-Islands model of economic behaviour. In particular, this is studied through the use of Agent-based Computational Economics, where individual economic agents are represented by separate computational entities in an interacting computer simulation. Three different economic models are studied: one where workers are assigned randomly to firms, the second where there is loyalty to firms from workers, and the third where workers have a broader set of criteria on which to make a job choice. Simulations show that there are positive relationships between level of rationality and several factors in the model, i.e. wage, vacancy rate and production, whilst unemployment level is negatively correlated with level of rationality.

Using semantic analysis, we present a technique known as semantically driven mutation which can explicitly detect and apply behavioural changes caused by the syntactic changes in programs that result from the mutation operation. Using semantically driven mutation, we demonstrate increased performance in genetic programming on seven benchmark genetic programming problems over two different domains.

Most real-world classification problems involve continuous (real-valued) attributes, as well as, nominal (discrete) attributes. The majority of ant colony optimisation (ACO) classification algorithms have the limitation of only being able to cope with nominal attributes directly. Extending the approach for coping with continuous attributes presented by cAnt-Miner (Ant-Miner coping with continuous attributes), in this paper we propose two new methods for handling continuous attributes in ACO classification algorithms. The first method allows a more flexible representation of continuous attributes' intervals. The second method explores the problem of attribute interaction, which originates from the way that continuous attributes are handled in cAnt-Miner, in order to implement an improved pheromone updating method. Empirical evaluation on eight publicly available data sets shows that the proposed methods facilitate the discovery of more accurate classification models.

This paper presents an agent-based computational economics model (ACE) to study demand-pull and cost-push inflation. Moreover, it studies the effect of different levels of rationality on the equilibrium price and unemployment rate. The model examines three different economies. In the first economy workers choose firms randomly, in the second economy there is loyalty between workers and firms. In the last scenario workers are persistence to find jobs. Simulations show that there is a positive relationship between equilibrium price and level of rationality while there is a negative relationship with unemployment rate. Moreover, the model is able to reproduce the behaviour of demand-pull inflation and cost-push inflation without homogeneous and perfectly rational agents assumptions.

Crossover forms one of the core operations in genetic programming and has been the subject of many different investigations. We present a novel technique, based on semantic analysis of programs, which forces each crossover to make candidate programs take a new step in the behavioural search space. We demonstrate how this technique results in better performance and smaller solutions in two separate genetic programming experiments.

This paper explores the idea ofneutralityin heuristic optimization algorithms. In particular, the effect of having multiple levels of neutrality in representations is explored. Two experiments using afitness-adaptivewalk algorithm are carried out: the first is concerned with function optimization with random Boolean networks, the second with a tunable neutral mapping applied to the hierarchical if-and-only-if function. In both of these cases it is shown that a two-level neutral mapping can be found that performs better than both non-neutral mappings and mappings with a single level of neutrality.

MOTIVATION: Cellular processes often hinge upon specific interactions among proteins, and knowledge of these processes at a system level constitutes a major goal of proteomics. In particular, a greater understanding of protein-protein interactions can be gained via a more detailed investigation of the protein domain interactions that mediate the interactions of proteins. Existing high-throughput experimental techniques assay protein-protein interactions, yet they do not provide any direct information on the interactions among domains. Inferences concerning the latter can be made by analysis of the domain composition of a set of proteins and their interaction map. This inference problem is non-trivial, however, due to the high level of noise generally present in experimental data concerning protein-protein interactions. This noise leads to contradictions, i.e. the impossibility of having a pattern of domain interactions compatible with the protein-protein interaction map. RESULTS: We formulate the problem of prediction of protein domain interactions in a form that lends itself to the application of belief propagation, a powerful algorithm for such inference problems, which is based on message passing. The input to our algorithm is an interaction map among a set of proteins, and a set of domain assignments to the relevant proteins. The output is a list of probabilities of interaction between each pair of domains. Our method is able to effectively cope with errors in the protein-protein interaction dataset and systematically resolve contradictions. We applied the method to a dataset concerning the budding yeast Saccharomyces cerevisiae and tested the quality of our predictions by cross-validation on this dataset, by comparison with existing computational predictions, and finally with experimentally available domain interactions. Results compare favourably to those by existing algorithms. AVAILABILITY: A C language implementation of the algorithm is available upon request.

This paper describes a genetic algorithm approach to coverage problems, that is, problems where the aim is to discover an example for each class in a given classification scheme.

Many data mining applications involve the task of building a model for predictive classification. The goal of such a model is to classify examples (records or data instances) into classes or categories of the same type. The use of variables (attributes) not related to the classes can reduce the accuracy and reliability of a classification or prediction model. Superuous variables can also increase the costs of building a model - particularly on large data sets. We propose a discrete Particle Swarm Optimization (PSO) algorithm designed for attribute selection. The proposed algorithm deals with discrete variables, and its population of candidate solutions contains particles of different sizes. The performance of this algorithm is compared with the performance of a standard binary PSO algorithm on the task of selecting attributes in a bioinformatics data set. The criteria used for comparison are: (1) maximizing predictive accuracy; and (2) finding the smallest subset of attributes.

We introduce a novel technique for the visualization of data at various levels of detail. This is based on a colour-based representation of the data, where "high level" views of the data are obtained by merging colours together to obtain a summary-colour which represents a number of data-points. This is applied to the problem of visualizing biomolecular sequence data and picking out features in such data at various scales.

Presents the mathematical and computational foundations of a robot programming environment embedded within a CAD system. The key ideas behind this system is that it will work offline, it will allow a high-level of task abstraction and it will be usable by designers and engineers who have a good knowledge of the desired task but only a basic grounding in robot engineering. We begin with a discussion of how robot workspace can be modelled using free-form CAD design concepts. The core of the paper is concerned with the application of these to well known problems of collision detection and path planning, showing how algorithms developed in CAD can be applied to these new problem areas in an efficient way. We use these ideas to consider the development of new, graphically-based, robot programming systems.