Citation Context

...is important that the Big Data mining techniques should be able to adapt and in some cases to detect change first. For example, the data stream mining field has very powerful techniques for this task =-=[13]-=-. • Compression: Dealing with Big Data, the quantity of space needed to store it is very relevant. There are two main approaches: compression where we don’t loose anything, or sampling where we choose...

Citation Context

...gorithms work in dynamic environments, where data are continuously generated. Sensor networks, monitoring, traffic management, telecommunication, or web log analysis are examples of such applications =-=[1]-=-. In these dynamic environments, incoming data form a data stream characterized by huge volumes of instances and rapid arrival-rate which often requires quick, real-time response. Compared to static e...

Citation Context

...classifiers. On the other hand, a new type of problems is becoming more visible, one in which learning algorithms work in dynamic environments with data continuously generated in the form of a stream =-=[1]-=-. Processing data streams implies new requirements concerning limited amount of memory, small processing time, and one scan of incoming data. Moreover, the data distributions and definitions of target...

Citation Context

...er quality control mechanisms may not be reflected in this document. Changes may have been made to this work since it was submitted for publication. DOI: http://dx.doi.org/10.1016/j.ins.2013.12.011 mechanisms in order to adjust to changing environments. Moreover, depending on the rate of these changes, concept drifts are usually divided into sudden or gradual ones, both of which require different reactions [34]. As standard data mining algorithms are not capable of dealing with concept drifts and rigorous processing requirements posed by data streams, several new techniques have been proposed [14, 24]. Out of many algorithms proposed to tackle evolving data streams, ensemble methods play an important role. Due to their modularity, they provide a natural way of adapting to change by modifying their structure, either by retraining ensemble members, replacing old component classifiers with new ones, or updating rules for aggregating component predictions [23]. Current adaptive ensembles can be further divided into block-based and online approaches [14]. Block-based approaches are designed to work in environments were examples arrive in portions, called blocks or chunks. Most block ensembles p...

Citation Context

...and DSMS was discussed in the introduction. Here we focus on data stream mining algorithms. On-line data stream mining has been the focus of many research efforts, and a recent review can be found in =-=[22]-=-. For instance, Ester et al. [15] proposed extending a static clustering algorithm, namely DBScan, for continuous clustering of data streams. Similarly, there have been efforts to build online classif...

Citation Context

...orks, measurements in network monitoring and traffic management, log records or click-streams in web exploring, manufacturing processes, call detail records, email, blogging, twitter posts and others =-=[17]-=-. In fact, all data generated can be considered as streaming data or as a snapshot of streaming data, since it is obtained from an interval of time. In the data stream model, data arrive at high speed...

Citation Context

...) stream “concepts” (i.e. models generating data) are assumed to be stationary or evolving [15]. The task of mining data streams is typically focused on specific types of algorithms [16], [15], [13], =-=[17]-=-. In particular, techniques have been developed for clustering, classification, frequency counting, time series analysis, and changes diagnosis (concept drift detection). The remainder of this paper i...

Citation Context

...tationary, i.e., the target concept may change over time. Concept drift means that the concept about which data is being collected may shift from time to time, each time after some minimum permanence =-=[6]-=-. In this paper we address the detection and analysis of concept drift in financial markets by employing a methodology based on Artificial Neural Networks (ANN). ANN are a set of biologically inspired...

Citation Context

...e popularity of the items themselves, fresh-based recommendation approaches care for suggesting items taking mainly into account recent and novel user preferences. Driven by the work in stream mining =-=[10]-=-, we use different types of aging mechanisms to define the way that the historical information (in form of ratings) is incorporated in the recommendation process. Aging in streams is typically impleme...

Citation Context

...es in the derived models and as a result of high data volumes it might no longer be possible to process the data efficiently more than once. In addition to this edited book, Gama’s comprehensive book =-=[7]-=- focuses on a number of stream mining solutions based on adaptive learning algorithms. Thereby the set of examples is not only incremented for a given learning algorithm, but also outdated examples ar...