A evolução de redes neurais artificiais encontra aplicações em diversos campos na área de aprendizado de máquina, em particular, simulações de vida artificial onde uma população de indivíduos controlados por redes neurais se adaptam num ambiente virtual a fim de realizar uma determinada tarefa. Similar ao processo natural pelo qual o comportamento do organismo se modifica filogeneticamente através da complexificação do sistema nervoso, tais simulações oferecem uma nova abordagem sintética no estudo da inteligência, em contraposição aos métodos simbólicos tradicionais. Um recente método, conhecido por NEAT (NeuroEvolution of Augmenting Topologies), é capaz de obter os pesos e a própria topologia de rede neural utilizando algoritmos genéticos. A codificação utilizada pelo NEAT é flexível o suficiente para permitir evolução aberta e arquiteturas neurais arbitrárias. Este trabalho apresenta uma implementação do NEAT que pode ser utilizada em conjunto com um simulador de propósito geral, chamado Breve, formando uma plataforma para experimentos de vida artificial. A implementação proposta também estende o NEAT para lidar com redes neurais dinâmicas, onde o nível de ativação dos neurônios varia continuamente no tempo. Este novo modelo é comparado com o método tradicional numa tarefa clássica de controle não-supervisionado, mostrando um aumento de eficiência na busca pela solução do problema. Os resultados obtidos motivam o uso desta plataforma para experimentos de vida artificial, onde uma população de indivíduos interage continuamente com um ambiente dinâmico, se adaptando ao longo das gerações. / The evolution of artificial neural networks has a wide range of applicability in diverse areas in the field of machine learning, particularly, in artificial life simulations where a population of individuals, controlled by neural networks, adapts in a virtual environment in order to solve a given task. Resembling the natural process in which an organism\'s behavior is subjected to phylogenetic modifications through the complexification of the nervous system, such simulations offer a new synthetic approach in the investigation of intelligence, counter posing traditional symbolic methods. A recent method known as NEAT (NeuroEvolution of Augmenting Topologies), is able to obtain the synaptic weights and the topology with the aid of genetic algorithms. The encoding used by NEAT is flexible enough to allow for open-ended evolution and arbitrary neural architectures. This work presents a NEAT implementation especially suitable to be used with a general purpose simulator known as Breve, constituting a framework for artificial life experiments. The proposed implementation extends NEAT to include dynamical neuron models, where their inner state continuously varies over time. The new model is then compared to the traditional method in a classic unsupervised control benchmark task, showing an efficiency increase while solving the problem. The obtained results motivate the proposed framework for general experiments in artificial life, in which a population of individuals continuously interact with a dynamical environment, adapting through generations.

In every learning or training environment, exercises are the basis for practical learning. Learners need to practice in order to acquire new abilities and perfect those gained previously. However, not every exercise is valid for every learner: learners require exercises that match their ability levels. Hence, difficulty of an exercise could be defined as the amount of effort that a learner requires to successfully complete the exercise (its learning cost). Too high difficulties tend to discourage learners and make them drop out, whereas too low difficulties are perceived as unchallenging, resulting in loss of interest. Correctly estimating difficulties is hard and error-prone problem that tends to be done manually using domain-expert knowledge. Underestimating or overestimating difficulty generates a problem for learners, increasing dropout rates in learning environments. This paper presents a novel approach to improve difficulty estimations by using Neuroevolution. The method is based on measuring the computational cost that Neuroevolution algorithms require to successfully complete a given exercise and establishing similarities with previously gathered information from learners. For specific experiments presented, a game called PLMan has been used. PLMan is a PacMan-like game in which users have to program the Artificial Intelligence of the main character using a Prolog knowledge base. Results show that there exists a correlation between students’ learning costs and those of Neuroevolution. This suggests that the approach is valid, and measured difficulty of Neuroevolution algorithms may be used as estimation for student's difficulty in the proposed environment.

While evolutionary computation is well suited for automatic discovery in engineering, it can also be used to gain insight into how humans and organizations could perform more effectively. Using a real-world problem of innovation search in organizations as the motivating example, this dissertation formalizes human creative problem solving as competitive multi-agent search. It differs from existing single-agent and team-search problems in that the agents interact through knowledge of other agents' searches and through the dynamic changes in the search landscape caused by these searches. The main hypothesis is that evolutionary computation can be used to discover effective strategies for competitive multi-agent search. This hypothesis is verified in experiments using an abstract domain based on the NK model, i.e. partially correlated and tunably rugged fitness landscapes, and a concrete domain in the form of a social innovation game. In both domains, different specialized strategies are evolved for each different competitive environment, and also strategies that generalize across environments. Strategies evolved in the abstract domain are more effective and more complex than hand-designed strategies and one based on traditional tree search. Using a novel spherical visualization of the fitness landscapes of the abstract domain, insight is gained about how successful strategies work, e.g. by tracking positive changes in the landscape. In the concrete game domain, human players were modeled using backpropagation, and used as opponents to create environments for evolution. Evolved strategies scored significantly higher than the human models by using a different proportion of actions, providing insights into how performance could be improved in social innovation domains. The work thus provides a possible framework for studying various human creative activities as competitive multi-agent search in the future. / text

Evolution of neural networks, or neuroevolution, has been a successful approach to many low-level control problems such as pole balancing, vehicle control, and collision warning. However, certain types of problems — such as those involving strategic decision-making — have remained difficult to solve. This dissertation proposes the hypothesis that such problems are difficult because they are fractured: The correct action varies discontinuously as the agent moves from state to state. To evaluate this hypothesis, a method for measuring fracture using the concept of function variation of optimal policies is proposed. This metric is used to evaluate a popular neuroevolution algorithm, NEAT, empirically on a set of fractured problems. The results show that (1) NEAT does not usually perform well on such problems, and (2) the reason is that NEAT does not usually generate local decision regions, which would be useful in constructing a fractured decision boundary. To address this issue, two neuroevolution algorithms that model local decision regions are proposed: RBF-NEAT, which biases structural search by adding basis-function nodes, and Cascade-NEAT, which constrains structural search by constructing cascaded topologies. These algorithms are compared to NEAT on a set of fractured problems, demonstrating that this approach can improve performance significantly. A meta-level algorithm, SNAP-NEAT, is then developed to combine the strengths of NEAT, RBF-NEAT, and Cascade-NEAT. An evaluation in a set of benchmark problems shows that it is possible to achieve good performance even when it is not known a priori whether a problem is fractured or not. A final empirical comparison of these methods demonstrates that they can scale up to real-world tasks like keepaway and half-field soccer. These results shed new light on why constructive neuroevolution algorithms have difficulty in certain domains and illustrate how bias and constraint can be used to improve performance. Thus, this dissertation shows how neuroevolution can be scaled up from learning low-level control to learning strategic decision-making problems. / text

This work focuses on automatization of neural network design via the so-called neuroevolution, which employs evolutionary algorithms to construct artificial neural networks or optimise their parameters. The goal of the project is to design and implement an evolutionary algorithm which can be used in the process of designing and optimizing topologies of convolutional neural networks. The effectiveness of the proposed framework was experimentally evaluated on tasks of image classification on datasets MNIST and CIFAR10 and compared with relevant solutions. The results showed that neuroevolution has a potential to successfully find accurate and effective convolutional neural network architectures.

The aim of this Master's thesis is to describe basic technics of evolutionary computing, convolutional neural networks (CNN), and automated design of neural networks using neuroevolution ( NAS - Neural Architecture Search ). NAS techniques are currently being researched more and more, as they speed up and simplify the lengthy and complicated process of designing artificial neural networks. These techniques are also able to search for unconventional architectures that would not be found by classic methods. The work also contains the design and implementation of software capable of automated development of convolutional neural networks using the open-source library TensorFlow. The program uses a multiobjective NSGA-II algorithm for designing accurate and compact CNNs.

No description available.

Teaching simulated biped robots to walk is a popular problem in machine learning. However, until this thesis, evolving a biped controller has not been attempted through an indirect encoding, i.e. a compressed representation of the solution, despite the fact that natural bipeds such as humans evolved through such an indirect encoding (i.e. DNA). Thus the promise for indirect encoding is to evolve gaits that rival those seen in nature. In this thesis, an indirect encoding called HyperNEAT evolves a controller for a biped robot in a computer simulation. To most effectively explore the deceptive behavior space of biped walkers, novelty search is applied as a fitness metric. The result is that although the indirect encoding can evolve a stable bipedal gait, the overall neural architecture is brittle to small mutations. This result suggests that some capabilities might be necessary to include beyond indirect encoding, such as lifetime adaptation. Thus this thesis provides fresh insight into the requisite ingredients for the eventual achievement of fluid bipedal walking through artificial evolution.

An ambitious long-term goal for neuroevolution, which studies how artificial evolutionary processes can be driven to produce brain-like structures, is to evolve neurocontrollers with a high density of neurons and connections that can adapt and learn from past experience. Yet while neuroevolution has produced successful results in a variety of domains, the scale of natural brains remains far beyond reach. In this dissertation two extensions to the recently introduced Hypercube-based NeuroEvolution of Augmenting Topologies (HyperNEAT) approach are presented that are a step towards more brain-like artificial neural networks (ANNs). First, HyperNEAT is extended to evolve plastic ANNs that can learn from past experience. This new approach, called adaptive HyperNEAT, allows not only patterns of weights across the connectivity of an ANN to be generated by a function of its geometry, but also patterns of arbitrary local learning rules. Second, evolvable-substrate HyperNEAT (ES-HyperNEAT) is introduced, which relieves the user from deciding where the hidden nodes should be placed in a geometry that is potentially infinitely dense. This approach not only can evolve the location of every neuron in the network, but also can represent regions of varying density, which means resolution can increase holistically over evolution. The combined approach, adaptive ES-HyperNEAT, unifies for the first time in neuroevolution the abilities to indirectly encode connectivity through geometry, generate patterns of heterogeneous plasticity, and simultaneously encode the density and placement of nodes in space. The dissertation culminates in a major application domain that takes a step towards the general goal of adaptive neurocontrollers for legged locomotion.

This thesis investigates the use of Bayesian analysis upon an opponent¿s behaviour in order to determine the desired goals or strategy used by a given adversary. A terrain analysis approach utilising the A* algorithm is investigated, where a probability distribution between discrete behaviours of an opponent relative to a set of possible goals is generated. The Bayesian analysis of agent behaviour accurately determines the intended goal of an opponent agent, even when the opponent¿s actions are altered randomly. The environment of Poker is introduced and abstracted for ease of analysis. Bayes¿ theorem is used to generate an effective opponent model, categorizing behaviour according to its similarity with known styles of opponent. The accuracy of Bayes¿ rule yields a notable improvement in the performance of an agent once an opponent¿s style is understood. A hybrid of the Bayesian style predictor and a neuroevolutionary approach is shown to lead to effective dynamic play, in comparison to agents that do not use an opponent model. The use of recurrence in evolved networks is also shown to improve the performance and generalizability of an agent in a multiplayer environment. These strategies are then employed in the full-scale environment of Texas Hold¿em, where a betting round-based approach proves useful in determining and counteracting an opponent¿s play. It is shown that the use of opponent models, with the adaptive benefits of neuroevolution aid the performance of an agent, even when the behaviour of an opponent does not necessarily fit within the strict definitions of opponent ¿style¿. / Engineering and Physical Sciences Research Council (EPSRC)