Ensemble techniques are a powerful method for recognising and reacting to changes in non-stationary data. However, most researches into dynamic classification with ensembles assume that the true class label of each incoming point is available or easily obtained. This is unrealistic in most practical applications, especially in high-velocity streams where manually labeling each point is prohibitively expensive. To address this challenge, this paper proposes an algorithm, named Clustering and One-Class Classification Ensemble Learning (COCEL), which incorporates a stream clustering algorithm and an ensemble of one-class classifiers with active learning, for classification in dynamic data streams. The method exploits the intuitive relationship between clusters and one-class classifiers to cope with a small training set (or no training set) and improve with experience, self-modifying its internal state to cope with changes in the data stream. The proposed method is evaluated on synthetic data streams exhibiting concept evolution and concept drift and a collection of high-velocity real data streams where manually labeling each incoming point is infeasible or expensive and labor intensive. Finally, a comparative evaluation with peer stream classification ensembles shows that COCEL can achieve superior or comparative accuracy while typically requiring less than 0.01% of the stream labels.

Data stream clustering provides valuable insights into the evolving patterns of long sequences of continuously generated data objects. Most existing clustering methods focus on single-view data streams. In this paper, we propose a multi-view representation learning (MVRL) method for multi-view clustering of data streams. We first introduce an integrated representation learning model to learn a fused sparse affinity matrix across multiple views for spectral clustering. Motivated by the optimization procedure of the integrated representation learning model, we propose three consecutive stages: collaborative representation, the construction of individual global affinity matrices using a mapping function, and the calculation of a fused sparse affinity matrix using Euclidean projection. These stages allow the effective capture of the global and local structures of high-dimensional data objects. Moreover, each stage has a closed-form solution, which determines the upper bound of the computational cost and memory consumption. We then employ the construction residuals of the collaborative representation to adaptively update a dynamic set, which is used to preserve the representative data objects. The dynamic set efficiently transfers previously learned useful knowledge to the arriving data objects. Extensive experimental results on multi-view data stream datasets demonstrate the effectiveness of the proposed MVRL method.

Population-based evolutionary algorithms have great potential to handle multiobjective optimisation problems . However, the performance of these algorithms depends largely on problem characteristics. There is a need to improve these algorithms for wide applicability. References, often specified by the decision maker’s preference in different forms, are very effective to boost the performance of algorithms. This paper proposes a novel framework for effective use of references to strengthen algorithms. This framework considers references as search targets which can be adjusted based on the information collected during the search. The proposed framework is combined with new strategies, such as reference adaptation and adaptive local mating, to solve different types of problems. The proposed algorithm is compared with state-of-the-arts on a wide range of problems with diverse characteristics. The comparison and extensive sensitivity analysis demonstrate that the proposed algorithm is competitive and robust across different types of problems studied in this paper.

Environmental changes are traditionally considered intrinsic in evolutionary dynamic optimization. However, by ignoring that changes can instead be induced, we are ignoring that environmental changes can be eventually beneficial. To investigate the impact of artificial changes on the optimization speed up, we propose a framework for inducing artificial changes in any pseudo-Boolean or continuous optimization in this paper. Seven types of changes can be induced. Knowing when and how the changes occur allows us to design new strategies for evolutionary algorithms. Through computational experiments and illustrative examples, the impact of introducing changes in the optimization process is investigated. Experimental results indicate that changing the environments according to the proposed framework can lead to higher speed up, but not for all problems and change types. The best performance was obtained by change types that introduce plateaus and/or modify the gradient of regions of the fitness landscape around the current best solution. By doing this, the evolutionary dynamics is modified, eventually allowing the population to escape faster from local optima and reach new zones of the fitness landscape. Given a pseudo-Boolean or continuous optimization static problem, the proposed framework can be used to dynamically change the problem to speed up the optimization.

In this work, discrete dynamic optimization problems (DOPs) are theoretically \nanalysed according to the modifications produced in the fitness landscape during the optimization process. Using the proposed analysis framework, the following DOPs are analysed: problems generated by the XOR DOP generator, three versions of the dynamic 0-1 knapsack problem, one problem involving evolutionary robots in dynamic environments, and the random dynamics NK-model. The XOR DOP generator creates benchmark DOPs from any binary static optimization problem, which allows to explore the properties of the static problem in a dynamic environment. Three types of transformations occurring in the fitness landscapes are observed in the DOPs analysed here. They are caused by: i) permutation of solutions in the search space; ii) duplication of solutions; and iii) adding deviations to the fitness of a subset of solutions. The XOR DOP generator creates a special type of permutation that is not found in the other investigated DOPs. In this way, a new benchmark problem generator is proposed here based on the analysis performed, allowing to produce DOPs with six types of fitness landscape transformations, including those similar to the problems investigated in this paper. When compared to the XOR DOP generator, new algorithms can be tested and compared in a wider range of dynamic environments using the new generator. It is important to observe that some of the fitness transformations analysed here, like those caused by the duplication of solutions, are not currently explored in the evolutionary dynamic optimization area.