arXiv subjects

Emma Hart

Publications and source records attributed to Emma Hart.

28 records · Page 2Linked to original sources

arXiv2023

To Switch or not to Switch: Predicting the Benefit of Switching between Algorithms based on Trajectory Features

Dynamic algorithm selection aims to exploit the complementarity of multiple optimization algorithms by switching between them during the search. While these kinds of dynamic algorithms have been shown to have potential to outperform their component algorithms, it is still unclear how this potential can best be realized. One promising approach is to make use of landscape features to enable a per-run trajectory-based switch. Here, the samples seen by the first algorithm are used to create a set of features which describe the landscape from the perspective of the algorithm. These features are then used to predict what algorithm to switch to. In this work, we extend this per-run trajectory-based approach to consider a wide variety of potential points at which to perform the switch. We show that using a sliding window to capture the local landscape features contains information which can be used to predict whether a switch at that point would be beneficial to future performance. By analyzing the resulting models, we identify what features are most important to these predictions. Finally, by evaluating the importance of features and comparing these values between multiple algorithms, we show clear differences in the way the second algorithm interacts with the local landscape features found before the switch.

cs.AI↗

arXiv2022

Automated Algorithm Selection: from Feature-Based to Feature-Free Approaches

We propose a novel technique for algorithm-selection, applicable to optimisation domains in which there is implicit sequential information encapsulated in the data, e.g., in online bin-packing. Specifically we train two types of recurrent neural networks to predict a packing heuristic in online bin-packing, selecting from four well-known heuristics. As input, the RNN methods only use the sequence of item-sizes. This contrasts to typical approaches to algorithm-selection which require a model to be trained using domain-specific instance features that need to be first derived from the input data. The RNN approaches are shown to be capable of achieving within 5% of the oracle performance on between 80.88% to 97.63% of the instances, depending on the dataset. They are also shown to outperform classical machine learning models trained using derived features. Finally, we hypothesise that the proposed methods perform well when the instances exhibit some implicit structure that results in discriminatory performance with respect to a set of heuristics. We test this hypothesis by generating fourteen new datasets with increasing levels of structure, and show that there is a critical threshold of structure required before algorithm-selection delivers benefit.

cs.LG↗

arXiv2022

Augmenting Novelty Search with a Surrogate Model to Engineer Meta-Diversity in Ensembles of Classifiers

Using Neuroevolution combined with Novelty Search to promote behavioural diversity is capable of constructing high-performing ensembles for classification. However, using gradient descent to train evolved architectures during the search can be computationally prohibitive. Here we propose a method to overcome this limitation by using a surrogate model which estimates the behavioural distance between two neural network architectures required to calculate the sparseness term in Novelty Search. We demonstrate a speedup of 10 times over previous work and significantly improve on previous reported results on three benchmark datasets from Computer Vision -- CIFAR-10, CIFAR-100, and SVHN. This results from the expanded architecture search space facilitated by using a surrogate. Our method represents an improved paradigm for implementing horizontal scaling of learning algorithms by making an explicit search for diversity considerably more tractable for the same bounded resources.

cs.LG↗

arXiv2021

Comparison of atlas-based and neural-network-based semantic segmentation for DENSE MRI images

Two segmentation methods, one atlas-based and one neural-network-based, were compared to see how well they can each automatically segment the brain stem and cerebellum in Displacement Encoding with Stimulated Echoes Magnetic Resonance Imaging (DENSE-MRI) data. The segmentation is a pre-requisite for estimating the average displacements in these regions, which have recently been proposed as biomarkers in the diagnosis of Chiari Malformation type I (CMI). In numerical experiments, the segmentations of both methods were similar to manual segmentations provided by trained experts. It was found that, overall, the neural-network-based method alone produced more accurate segmentations than the atlas-based method did alone, but that a combination of the two methods -- in which the atlas-based method is used for the segmentation of the brain stem and the neural-network is used for the segmentation of the cerebellum -- may be the most successful.

eess.IV↗

arXiv2021

Morpho-evolution with learning using a controller archive as an inheritance mechanism

The joint optimisation of body-plan and control via evolutionary processes can be challenging in rich morphological spaces in which offspring can have body-plans that are very different from either of their parents. This causes a potential mismatch between the structure of an inherited controller and the new body. To address this, we propose a framework that combines an evolutionary algorithm to generate body-plans and a learning algorithm to optimise the parameters of a neural controller. The topology of this controller is created once the body-plan of each offspring body-plan is generated. The key novelty of the approach is to add an external archive for storing learned controllers that map to explicit `types' of robots (where this is defined with respect the features of the body-plan). By learning from a controller with an appropriate structure inherited from the archive, rather than from a randomly initialised one, we show that both the speed and magnitude of learning increases over time when compared to an approach that starts from scratch, using two tasks and three environments. The framework also provides new insights into the complex interactions between evolution and learning.

cs.RO↗

arXiv2019

Being a leader or being the leader: The evolution of institutionalised hierarchy

Human social hierarchy has the unique characteristic of existing in two forms. Firstly, as an informal hierarchy where leaders and followers are implicitly defined by their personal characteristics, and secondly, as an institutional hierarchy where leaders and followers are explicitly appointed by group decision. Although both forms can reduce the time spent in organising collective tasks, institutional hierarchy imposes additional costs. It is therefore natural to question why it emerges at all. The key difference lies in the fact that institutions can create hierarchy with only a single leader, which is unlikely to occur in unregulated informal hierarchy. To investigate if this difference can affect group decision-making and explain the evolution of institutional hierarchy, we first build an opinion-formation model that simulates group decision making. We show that in comparison to informal hierarchy, a single-leader hierarchy reduces (i) the time a group spends to reach consensus, (ii) the variation in consensus time, and (iii) the rate of increase in consensus time as group size increases. We then use this model to simulate the cost of organising a collective action which produces resources, and integrate this into an evolutionary model where individuals can choose between informal or institutional hierarchy. Our results demonstrate that groups evolve preferences towards institutional hierarchy, despite the cost of creating an institution, as it provides a greater organisational advantage which is less affected by group size and inequality.

q-bio.PE↗

arXiv2018

Can justice be fair when it is blind? How social network structures can promote or prevent the evolution of despotism

Hierarchy is an efficient way for a group to organize, but often goes along with inequality that benefits leaders. To control despotic behaviour, followers can assess leaders decisions by aggregating their own and their neighbours experience, and in response challenge despotic leaders. But in hierarchical social networks, this interactional justice can be limited by (i) the high influence of a small clique who are treated better, and (ii) the low connectedness of followers. Here we study how the connectedness of a social network affects the co-evolution of despotism in leaders and tolerance to despotism in followers. We simulate the evolution of a population of agents, where the influence of an agent is its number of social links. Whether a leader remains in power is controlled by the overall satisfaction of group members, as determined by their joint assessment of the leaders behaviour. We demonstrate that centralization of a social network around a highly influential clique greatly increases the level of despotism. This is because the clique is more satisfied, and their higher influence spreads their positive opinion of the leader throughout the network. Finally, our results suggest that increasing the connectedness of followers limits despotism while maintaining hierarchy.

q-bio.PE↗

arXiv2018

Optimisation and Illumination of a Real-world Workforce Scheduling and Routing Application via Map-Elites

Workforce Scheduling and Routing Problems (WSRP) are very common in many practical domains, and usually, have a number of objectives. Illumination algorithms such as Map-Elites (ME) have recently gained traction in application to {\em design} problems, in providing multiple diverse solutions as well as illuminating the solution space in terms of user-defined characteristics, but typically require significant computational effort to produce the solution archive. We investigate whether ME can provide an effective approach to solving WSRP, a {\em repetitive} problem in which solutions have to be produced quickly and often. The goals of the paper are two-fold. The first is to evaluate whether ME can provide solutions of competitive quality to an Evolutionary Algorithm (EA) in terms of a single objective function, and the second to examine its ability to provide a repertoire of solutions that maximise user choice. We find that very small computational budgets favour the EA in terms of quality, but ME outperforms the EA at larger budgets, provides a more diverse array of solutions, and lends insight to the end-user.

cs.AI↗

arXiv2018

An Investigation of Environmental Influence on the Benefits of Adaptation Mechanisms in Evolutionary Swarm Robotics

A robotic swarm that is required to operate for long periods in a potentially unknown environment can use both evolution and individual learning methods in order to adapt. However, the role played by the environment in influencing the effectiveness of each type of learning is not well understood. In this paper, we address this question by analysing the performance of a swarm in a range of simulated, dynamic environments where a distributed evolutionary algorithm for evolving a controller is augmented with a number of different individual learning mechanisms. The learning mechanisms themselves are defined by parameters which can be either fixed or inherited. We conduct experiments in a range of dynamic environments whose characteristics are varied so as to present different opportunities for learning. Results enable us to map environmental characteristics to the most effective learning algorithm.

cs.NE↗

arXiv2018

Evolution of a Functionally Diverse Swarm via a Novel Decentralised Quality-Diversity Algorithm

The presence of functional diversity within a group has been demonstrated to lead to greater robustness, higher performance and increased problem-solving ability in a broad range of studies that includes insect groups, human groups and swarm robotics. Evolving group diversity however has proved challenging within Evolutionary Robotics, requiring reproductive isolation and careful attention to population size and selection mechanisms. To tackle this issue, we introduce a novel, decentralised, variant of the MAP-Elites illumination algorithm which is hybridised with a well-known distributed evolutionary algorithm (mEDEA). The algorithm simultaneously evolves multiple diverse behaviours for multiple robots, with respect to a simple token-gathering task. Each robot in the swarm maintains a local archive defined by two pre-specified functional traits which is shared with robots it come into contact with. We investigate four different strategies for sharing, exploiting and combining local archives and compare results to mEDEA. Experimental results show that in contrast to previous claims, it is possible to evolve a functionally diverse swarm without geographical isolation, and that the new method outperforms mEDEA in terms of the diversity, coverage and precision of the evolved swarm.

cs.NE↗