Unsupervised Anomaly Detection in Stream Data with Online Evolving Spiking Neural Networks

Unsupervised anomaly discovery in stream data is a research topic with many practical applications. However, in many cases, it is not easy to collect enough training data with labeled anomalies for supervised learning of an anomaly detector in order to deploy it later for identification of real anomalies in streaming data. It is thus important to design anomalies detectors that can correctly detect anomalies without access to labeled training data. Our idea is to adapt the Online evolving Spiking Neural Network (OeSNN) classifier to the anomaly detection task. As a result, we offer an Online evolving Spiking Neural Network for Unsupervised Anomaly Detection algorithm (OeSNN-UAD), which, unlike OeSNN, works in an unsupervised way and does not separate output neurons into disjoint decision classes. OeSNN-UAD uses our proposed new two-step anomaly detection method. Also, we derive new theoretical properties of neuronal model and input layer encoding of OeSNN, which enable more effective and efficient detection of anomalies in our OeSNN-UAD approach. The proposed OeSNN-UAD detector was experimentally compared with state-of-the-art unsupervised and semi-supervised detectors of anomalies in stream data from the Numenta Anomaly Benchmark and Yahoo Anomaly Datasets repositories. Our approach outperforms the other solutions provided in the literature in the case of data streams from the Numenta Anomaly Benchmark repository. Also, in the case of real data files of the Yahoo Anomaly Benchmark repository, OeSNN-UAD outperforms other selected algorithms, whereas in the case of Yahoo Anomaly Benchmark synthetic data files, it provides competitive results to the results recently reported in the literature.

From numbers to a biological sense: How the strategy chosen for metabolomics data treatment may affect final results. A practical example based on urine fingerprints obtained by LC-MS

Application of high-throughput technologies in metabolomics studies increases the quantity of data obtained, which in turn imposes several problems during data analysis. Correctly and clearly addressed biological question and comprehensive knowledge about data structure and properties are definitely necessary to select proper chemometric tools. However, there is a broad range of chemometric tools available for use with metabolomics data, which makes this choice challenging. Precisely performed data treatment enables valuable extraction of information and its proper interpretation. The effect of an error made at an early stage will be enhanced throughout the later stages, which in combination with other errors made at each step can accumulate and significantly affect the data interpretation. Moreover, adequate application of these tools may help not only to detect, but sometimes also to correct, biological, analytical, or methodological errors, which may affect truthfulness of obtained results. This report presents steps and tools used for LC-MS based metabolomics data extraction, reduction, and visualization. Following such steps as data reprocessing, data pretreatment, data treatment, and data revision, authors want to show how to extract valuable information and how to avoid misinterpretation of results obtained. The purpose of this work was to emphasize problematic characteristics of metabolomics data and the necessity for their attentive and precise treatment. The dataset used to illustrate metabolomics data properties and to illustrate major data treatment challenges was obtained utilizing an animal model of control and diabetic rats, both with and without rosemary treatment. Urine samples were fingerprinted employing LC-QTOF-MS.