Meta-World is widely used for evaluating multi-task and meta-reinforcement learning agents, which are challenged to master diverse skills simultaneously. Since its introduction however, there have been numerous undocumented changes which inhibit a fair comparison of algorithms. This work strives to disambiguate these results from the literature, while also leveraging the past versions of Meta-World to provide insights into multi-task and meta-reinforcement learning benchmark design. Through this process we release an open-source version of Meta-World that has full reproducibility of past results, is more technically ergonomic, and gives users more control over the tasks that are included in a task set.

The main focus of this thesis is to compare the ability of various swarm intelligence algorithms when applied to the training of artificial neural networks. In order to compare the performance of the selected swarm intelligence algorithms both classification and regression datasets were chosen from the UCI Machine Learning repository. Swarm intelligence algorithms are compared in terms of training loss, training accuracy, testing loss, testing accuracy, hidden unit saturation, and overfitting. Our observations showed that Particle Swarm Optimization (PSO) was the best performing algorithm in terms of Training loss and Training accuracy. However, it was also found that the performance of PSO dropped considerably when examining the testing loss and testing accuracy results. For the classification problems, it was found that firefly algorithm, ant colony optimization, and fish school search outperformed PSO for testing loss and testing accuracy. It was also observed that ant colony optimization was the algorithm that performed the best in terms of hidden unit saturation.