Hyperparameter optimization is crucial for maximizing machine learning model performance, yet most existing algorithms are designed for batch or offline scenarios and assume static data distributions.Such assumptions fall short in data stream settings, where models must adapt to evolving inputs in real time.To address these limitations, we propose the Bayesian Stream Tuner (BST), a novel framework for online hyperparameter optimization in nonstationary data streams.BST maintains a dynamic set of candidate hyperparameter configurations and periodically refines them using an incremental Bayesian model, which estimates configuration performance based on recent data statistics and hyperparameter values.This systematic exploration and refinement strategy allows BST to detect and respond to concept drift by resetting its adaptation mechanisms whenever necessary, ensuring strong performance under changing distributions.Our theoretical analysis establishes sublinear regret bounds for BST in dynamic environments, and extensive experiments on classification and regression tasks demonstrate that BST consistently outperforms state-of-the-art online hyperparameter optimization methods in both predictive accuracy and adaptability, making it a powerful solution for real-time hyperparameter tuning in evolving data streams.

Automated Machine Learning (autoML) is a novel topic that aims to tackle the parameter configuration issue using automatic monitoring models and comprises different machine learning tasks, such as feature selection, model selection, and hyper-parameter tuning. It makes easier use of algorithms for non-ML experts as well as ML experts by automating tasks that rely on expert domain knowledge. Nevertheless, autoML is in its infancy stage and not well explored yet in the offline and stream settings. In this paper, we propose automated Classification (auto-Class) method for automated algorithm selection and configuration for data stream classification. AutoClass consists of training an ensemble of different tuned configurations and selecting the best-performing configuration to do the prediction. We present experiments performed on a diverse set of real and artificial datasets and show how our proposed approach can outperform the performance of competitive state-of-the-art ensemble and single-based methods.

Many real-world applications involve classification from evolving data streams. However, learning in such environment requires algorithms able to learn and predict from potentially unbounded data that are constantly changing. For this to happen, stream algorithms should restrict the storage to a part of – and/or synopsis information from – the stream using efficient and accurate manners and strategies, such as window models and summarization techniques (e.g., sampling, sketching, dimensionality reduction). In this work, we focus on the k-Nearest Neighbors (kNN) where most of the existing approaches for data streams consider that instances have the same weight from the start to the finish of the processing task.In a streaming data scenario, it is often the case that the most recent elements from the data stream are the more relevant ones. Taking into account that the most recent instances are more relevant, we propose a novel kNN approach that stores instances in a sliding window and weighs them according to their arrival time (i.e position on the window) using an adjusted weight function. The empirical results on comprehensive real and synthetic datasets indicate the effectiveness and efficiency of our proposed approach in comparison with state-of-the-art algorithms.

Over the last decade, we witnessed the proliferation of several machine learning algorithms capable of solving different tasks for the most diverse applications. Often, for an algorithm to be effective, significant human effort is required, in particular for hyper-parameter tuning and data cleaning. Recently, there have been increasing efforts to alleviate such a burden and make machine learning algorithms easier to use for researchers with varying levels of expertise. Nevertheless, the question of whether an efficient and fully generalizable automated Machine Learning (autoML) framework is possible remains unanswered. In this paper, we present autoAD, the first autoML framework for unsupervised anomaly detection. By leveraging a pool of different anomaly detection algorithms, each one coming with its own hyper-parameter search space, our framework automatically selects the best performing approach, while determining an optimal configuration for its hyper-parameters on a given dataset. Our extensive experimental evaluation, conducted on a rich collection of datasets, shows the substantial gains that can be achieved with autoAD compared to state-of-the-art methods for unsupervised anomaly detection.

The last few decades have witnessed a significant evolution of technology in different domains, changing the way the world operates, which leads to an overwhelming amount of data generated in an open-ended way as streams. Over the past years, we observed the development of several machine learning algorithms to process big data streams. However, the accuracy of these algorithms is very sensitive to their hyper-parameters, which requires expertise and extensive trials to tune. Another relevant aspect is the high-dimensionality of data, which can causes degradation to computational performance. To cope with these issues, this paper proposes a stream k-nearest neighbors (kNN) algorithm that applies an internal dimension reduction to the stream in order to reduce the resource usage and uses an automatic monitoring system that tunes dynamically the configuration of the kNN algorithm and the output dimension size with big data streams. Experiments over a wide range of datasets show that the predictive and computational performances of the kNN algorithm are improved.

Learning from potentially infinite and high-dimensional data streams poses significant challenges in the classification task. For instance, k -Nearest Neighbors ( k NN) is one of the most often used algorithms in the data stream mining area that proved to be very resource-intensive when dealing with high-dimensional spaces. Uniform Manifold Approximation and Projection (UMAP) is a novel manifold technique and one of the most promising dimension reduction and visualization techniques in the non-streaming setting because of its high performance in comparison with competitors. However, there is no version of UMAP that copes with the challenging context of streams. To overcome these restrictions, we propose a batch-incremental approach that pre-processes data streams using UMAP, by producing successive embeddings on a stream of disjoint batches in order to support an incremental k NN classification. Experiments conducted on publicly available synthetic and real-world datasets demonstrate the substantial gains that can be achieved with our proposal compared to state-of-the-art techniques.

A well-known learning task in big data stream mining is classification. Extensively studied in the offline setting, in the streaming setting - where data are evolving and even infinite - it is still a challenge. In the offline setting, training needs to store all the data in memory for the learning task; yet, in the streaming setting, this is impossible to do due to the massive amount of data that is generated in real-time. To cope with these resource issues, this paper proposes and analyzes several evolving naive Bayes classification algorithms, based on the well-known count-min sketch, in order to minimize the space needed to store the training data. The proposed algorithms also adapt concept drift approaches, such as ADWIN, to deal with the fact that streaming data may be evolving and change over time. However, handling sparse, very high-dimensional data in such framework is highly challenging. Therefore, we include the hashing trick, a technique for dimensionality reduction, to compress that down to a lower dimensional space, which leads to a large memory saving.We give a theoretical analysis which demonstrates that our proposed algorithms provide a similar accuracy quality to the classical big data stream mining algorithms using a reasonable amount of resources. We validate these theoretical results by an extensive evaluation on both synthetic and real-world datasets.