We look at how the structure of social networks and the nature of social interactions affect the ... more We look at how the structure of social networks and the nature of social interactions affect the behaviour of Particle Swarms Optimisers. To this end, we propose a general model of communication and consensus which focuses on the effects of social interactions putting the details of the dynamics and the optimum seeking behaviour of PSOs into the background.
Workshop Italiano di Vita Artificiale e …, Jan 1, 2007
We present a study concerning the use of artificial neural networks for object tracking/segmentat... more We present a study concerning the use of artificial neural networks for object tracking/segmentation in surveillance video sequences. A number of artificial creatures, conceptually, "inhabit" our image sequences. Their task is to explore the images looking for moving objects. Their behaviour is controlled by neural networks which are evolved with an Evolutionary Algorithm (EA). In our system we encourage the EA to evolve a population of individuals with behaviours which are as different as possible. In addition, we require populations to satisfy some minimal requirements necessary for the object tracking behaviour to emerge from the interaction of individuals. Population performance is evaluated on artificial video sequences and some results are discussed.
Fitness-Proportional Negative Slope Coefficient is a fitness landscapes measure that has recently... more Fitness-Proportional Negative Slope Coefficient is a fitness landscapes measure that has recently been introduced as a potential indicator of problem hardness for optimisation. It is inspired to an older measure, the Negative Slope Coefficient, and it has been theoretically modelled. Preliminary experiments have suggested that it may be a good predictor of problem hardness. However, this measure has not undergone any convincing and comprehensive empirical testing. Our objective is to fill this gap. So, we perform empirical tests using a large set of invertible functions of unitation. We find that while this measure may correctly predict the degree of evolvability of a landscape, this does not necessarily correlate with the difficulty of problems. Some landscapes may show, for example, limited evolvability and yet be easy to solve because either solutions are already present in the initial population or the computational resources provided exceed evolvability obstacles. Or it may be impossible to solve them irrespective of their evolvability simply because they are far too vast for the computational resources provided. These situations are hardly captured by the Fitness-Proportional Negative Slope Coefficient.
... like to thank EvoNet, the Network of Excellence in Evolutionary Computing, for its support, i... more ... like to thank EvoNet, the Network of Excellence in Evolutionary Computing, for its support, in particular to Jennifer Willies and ... 59 Gianluigi Folino, Clara Pizzuti, and Giandomenico Spezzano Improving Symbolic Regression with Interval Arithmetic and Linear Scaling 70 Maarten ...
The phenomenon of growth in program size in genetic programming populations has been widely repor... more The phenomenon of growth in program size in genetic programming populations has been widely reported. In a variety of experiments and static analysis we test the standard protective code explanation and find it to be incomplete. We suggest bloat is primarily due to distribution of fitness in the space of possible programs and because of this, in the absence of bias, it is in general inherent in any search technique using a variable length representation.
... of Karlsruhe, Karlsruhe, Germany. Riccardo Poli, Univ. ... The evolution and analysis of pote... more ... of Karlsruhe, Karlsruhe, Germany. Riccardo Poli, Univ. ... The evolution and analysis of potential antibody library for use in job-shop scheduling [full citation] Emma Hart, Peter Ross. Immune network modelling in design optimization [full citation] Prabhat Hajela, Jun Sun Yoo. ...
In this paper we look at how the structure of the social network and the nature of the social int... more In this paper we look at how the structure of the social network and the nature of the social interactions can affect the behaviour of particle swarms. To this end, we create and test a general model of communication and consensus which puts the details of the dynamics and the optimum seeking behaviour of PSOs into the background, thereby allowing us to focus on the effects of social interactions.
This paper presents two new schema theorems in which expectations are absent. The first theorem p... more This paper presents two new schema theorems in which expectations are absent. The first theorem provides confidence intervals for the number of instances of a schema at the next generation. The second gives a lower bound for the same quantity.
In previous work, we have introduced an effective and resource-efficient hyperheuristic that uses... more In previous work, we have introduced an effective and resource-efficient hyperheuristic that uses Genetic Programming as its search heuristic on the space of heuristics. Here, we show that the hyperheuristic performs better than purely greedy and even only mostly greedy flavours of hill climbing. We also introduce a generic principle that allows the hyperheuristic to automatically find good parameter values for its effective and efficient search.
The effects of neutrality on evolutionary search have been considered in a number of interesting ... more The effects of neutrality on evolutionary search have been considered in a number of interesting studies, the results of which, however, have been contradictory. We believe that this confusion is due to several reasons. In this paper, we shed some light on neutrality by addressing these problems. That is, we use the simplest possible definition of neutrality, we consider one of the simplest possible algorithms, we apply it to two problems (a unimodal landscape and a deceptive landscape), which we analyse using fitness distance correlation, performance statistics and, critically, tracking the full evolutionary path of individuals within their family tree.
Recent theoretical work has characterised the search bias of GP subtree swapping crossover in ter... more Recent theoretical work has characterised the search bias of GP subtree swapping crossover in terms of program length distributions, providing an exact fixed point for trees with internal nodes of identical arity. However, only an approximate model (based on the notion of average arity) for the mixed-arity case has been proposed. This leaves a particularly important gap in our knowledge because multi-arity function sets are commonplace in GP and deep lessons could be learnt from the fixed point. In this paper, we present an accurate theoretical model of program length distributions when mixed-arity function sets are employed. The new model is based on the notion of an arity histogram, a count of the number of primitives of each arity in a program. Empirical support is provided and a discussion of the model is used to place earlier findings into a more general context. 39 selection and a new method to control sampling, by length, of GP operators -operator equalisation.
We present, GP-HH, a framework for evolving local search 3-SAT heuristics based on GP. Evolved he... more We present, GP-HH, a framework for evolving local search 3-SAT heuristics based on GP. Evolved heuristics are compared against well-known SAT solvers with very encouraging results.
Metaheuristics have often been shown to be effective for difficult combinatorial optimization pro... more Metaheuristics have often been shown to be effective for difficult combinatorial optimization problems. The reason for that, however, remains unclear. A framework for a theory of metaheuristics crucially depends on a formal representative model of such algorithms. This paper unifies/reconciles in a single framework the model of a black box algorithm coming from the no-free-lunch research (e.g. Wolpert et al.
We provide strong evidence that sub-tree swapping crossover when applied to tree-based representa... more We provide strong evidence that sub-tree swapping crossover when applied to tree-based representations will cause alleles (node labels) to diffuse within length classes. For a-ary trees we provide further confirmation that all programs are equally likely to be sampled within any length class when sub-tree swapping crossover is applied in the absence of selection and mutation. Therefore, we propose that this form of search is unbiased -within length classes -for a-ary trees. Unexpectedly, however, for mixed-arity trees this is not found and a more complicated form of search is taking place where certain tree shapes, hence programs, are more likely to be sampled than others within each class. We examine the reasons for such shape bias in mixed arity representations and provide the practitioner with a thorough examination of sub-tree swapping crossover bias. The results of this, when combined with crossover length bias research, explain Genetic Programming's lack of structural convergence during later stages of an experimental run. Several operators are discussed where a broader form of convergence may be detected in a similar way to that found in Genetic Algorithm experimentation.
Our previous work has introduced a hyperheuristic (HH) approach based on Genetic Programming (GP)... more Our previous work has introduced a hyperheuristic (HH) approach based on Genetic Programming (GP). There, GP employs usergiven languages where domain-specific local heuristics are used as primitives for producing specialised metaheuristics (MH). Here, we show that the GP-HH works well with simple generic languages over subheuristic primitives, dealing with increases of problem size and reduction of resources. The system produces effective and efficient MHs that deliver best results known in a chosen test domain. We also demonstrate that user-given, modest domain information allows the HH to produce an improvement over a previous best result from the literature.
We look at how the structure of social networks and the nature of social interactions affect the ... more We look at how the structure of social networks and the nature of social interactions affect the behaviour of Particle Swarms Optimisers. To this end, we propose a general model of communication and consensus which focuses on the effects of social interactions putting the details of the dynamics and the optimum seeking behaviour of PSOs into the background.
Workshop Italiano di Vita Artificiale e …, Jan 1, 2007
We present a study concerning the use of artificial neural networks for object tracking/segmentat... more We present a study concerning the use of artificial neural networks for object tracking/segmentation in surveillance video sequences. A number of artificial creatures, conceptually, "inhabit" our image sequences. Their task is to explore the images looking for moving objects. Their behaviour is controlled by neural networks which are evolved with an Evolutionary Algorithm (EA). In our system we encourage the EA to evolve a population of individuals with behaviours which are as different as possible. In addition, we require populations to satisfy some minimal requirements necessary for the object tracking behaviour to emerge from the interaction of individuals. Population performance is evaluated on artificial video sequences and some results are discussed.
Fitness-Proportional Negative Slope Coefficient is a fitness landscapes measure that has recently... more Fitness-Proportional Negative Slope Coefficient is a fitness landscapes measure that has recently been introduced as a potential indicator of problem hardness for optimisation. It is inspired to an older measure, the Negative Slope Coefficient, and it has been theoretically modelled. Preliminary experiments have suggested that it may be a good predictor of problem hardness. However, this measure has not undergone any convincing and comprehensive empirical testing. Our objective is to fill this gap. So, we perform empirical tests using a large set of invertible functions of unitation. We find that while this measure may correctly predict the degree of evolvability of a landscape, this does not necessarily correlate with the difficulty of problems. Some landscapes may show, for example, limited evolvability and yet be easy to solve because either solutions are already present in the initial population or the computational resources provided exceed evolvability obstacles. Or it may be impossible to solve them irrespective of their evolvability simply because they are far too vast for the computational resources provided. These situations are hardly captured by the Fitness-Proportional Negative Slope Coefficient.
... like to thank EvoNet, the Network of Excellence in Evolutionary Computing, for its support, i... more ... like to thank EvoNet, the Network of Excellence in Evolutionary Computing, for its support, in particular to Jennifer Willies and ... 59 Gianluigi Folino, Clara Pizzuti, and Giandomenico Spezzano Improving Symbolic Regression with Interval Arithmetic and Linear Scaling 70 Maarten ...
The phenomenon of growth in program size in genetic programming populations has been widely repor... more The phenomenon of growth in program size in genetic programming populations has been widely reported. In a variety of experiments and static analysis we test the standard protective code explanation and find it to be incomplete. We suggest bloat is primarily due to distribution of fitness in the space of possible programs and because of this, in the absence of bias, it is in general inherent in any search technique using a variable length representation.
... of Karlsruhe, Karlsruhe, Germany. Riccardo Poli, Univ. ... The evolution and analysis of pote... more ... of Karlsruhe, Karlsruhe, Germany. Riccardo Poli, Univ. ... The evolution and analysis of potential antibody library for use in job-shop scheduling [full citation] Emma Hart, Peter Ross. Immune network modelling in design optimization [full citation] Prabhat Hajela, Jun Sun Yoo. ...
In this paper we look at how the structure of the social network and the nature of the social int... more In this paper we look at how the structure of the social network and the nature of the social interactions can affect the behaviour of particle swarms. To this end, we create and test a general model of communication and consensus which puts the details of the dynamics and the optimum seeking behaviour of PSOs into the background, thereby allowing us to focus on the effects of social interactions.
This paper presents two new schema theorems in which expectations are absent. The first theorem p... more This paper presents two new schema theorems in which expectations are absent. The first theorem provides confidence intervals for the number of instances of a schema at the next generation. The second gives a lower bound for the same quantity.
In previous work, we have introduced an effective and resource-efficient hyperheuristic that uses... more In previous work, we have introduced an effective and resource-efficient hyperheuristic that uses Genetic Programming as its search heuristic on the space of heuristics. Here, we show that the hyperheuristic performs better than purely greedy and even only mostly greedy flavours of hill climbing. We also introduce a generic principle that allows the hyperheuristic to automatically find good parameter values for its effective and efficient search.
The effects of neutrality on evolutionary search have been considered in a number of interesting ... more The effects of neutrality on evolutionary search have been considered in a number of interesting studies, the results of which, however, have been contradictory. We believe that this confusion is due to several reasons. In this paper, we shed some light on neutrality by addressing these problems. That is, we use the simplest possible definition of neutrality, we consider one of the simplest possible algorithms, we apply it to two problems (a unimodal landscape and a deceptive landscape), which we analyse using fitness distance correlation, performance statistics and, critically, tracking the full evolutionary path of individuals within their family tree.
Recent theoretical work has characterised the search bias of GP subtree swapping crossover in ter... more Recent theoretical work has characterised the search bias of GP subtree swapping crossover in terms of program length distributions, providing an exact fixed point for trees with internal nodes of identical arity. However, only an approximate model (based on the notion of average arity) for the mixed-arity case has been proposed. This leaves a particularly important gap in our knowledge because multi-arity function sets are commonplace in GP and deep lessons could be learnt from the fixed point. In this paper, we present an accurate theoretical model of program length distributions when mixed-arity function sets are employed. The new model is based on the notion of an arity histogram, a count of the number of primitives of each arity in a program. Empirical support is provided and a discussion of the model is used to place earlier findings into a more general context. 39 selection and a new method to control sampling, by length, of GP operators -operator equalisation.
We present, GP-HH, a framework for evolving local search 3-SAT heuristics based on GP. Evolved he... more We present, GP-HH, a framework for evolving local search 3-SAT heuristics based on GP. Evolved heuristics are compared against well-known SAT solvers with very encouraging results.
Metaheuristics have often been shown to be effective for difficult combinatorial optimization pro... more Metaheuristics have often been shown to be effective for difficult combinatorial optimization problems. The reason for that, however, remains unclear. A framework for a theory of metaheuristics crucially depends on a formal representative model of such algorithms. This paper unifies/reconciles in a single framework the model of a black box algorithm coming from the no-free-lunch research (e.g. Wolpert et al.
We provide strong evidence that sub-tree swapping crossover when applied to tree-based representa... more We provide strong evidence that sub-tree swapping crossover when applied to tree-based representations will cause alleles (node labels) to diffuse within length classes. For a-ary trees we provide further confirmation that all programs are equally likely to be sampled within any length class when sub-tree swapping crossover is applied in the absence of selection and mutation. Therefore, we propose that this form of search is unbiased -within length classes -for a-ary trees. Unexpectedly, however, for mixed-arity trees this is not found and a more complicated form of search is taking place where certain tree shapes, hence programs, are more likely to be sampled than others within each class. We examine the reasons for such shape bias in mixed arity representations and provide the practitioner with a thorough examination of sub-tree swapping crossover bias. The results of this, when combined with crossover length bias research, explain Genetic Programming's lack of structural convergence during later stages of an experimental run. Several operators are discussed where a broader form of convergence may be detected in a similar way to that found in Genetic Algorithm experimentation.
Our previous work has introduced a hyperheuristic (HH) approach based on Genetic Programming (GP)... more Our previous work has introduced a hyperheuristic (HH) approach based on Genetic Programming (GP). There, GP employs usergiven languages where domain-specific local heuristics are used as primitives for producing specialised metaheuristics (MH). Here, we show that the GP-HH works well with simple generic languages over subheuristic primitives, dealing with increases of problem size and reduction of resources. The system produces effective and efficient MHs that deliver best results known in a chosen test domain. We also demonstrate that user-given, modest domain information allows the HH to produce an improvement over a previous best result from the literature.
Uploads
Papers by Riccardo Poli