Evolutionary Rule Mining in Time Series Databases

Abstract

Abstract. Data mining in the form of rule discovery is a growing field of investigation. A recent addition to this field is the use of evolutionary algorithms in the mining process. While this has been used extensively in the traditional mining of relational databases, it has hardly, if at all, been used in mining sequences and time series. In this paper we describe our method for evolutionary sequence mining, using a specialized piece of hardware for rule evaluation, and show how the method can be applied to several different mining tasks, such as supervised sequence prediction, unsupervised mining of interesting rules, discovering connections between separate time series, and investigating tradeoffs between contradictory objectives by using multiobjective evolution.

An energy-efficient querying framework in sensor networks for detecting node similarities

“…The AR model provides a small and cheap fingerprint of the data stream produced at the sensor, and it performs well in presence of temporal fluctuating data (as we showed in [25].) There has also been a much work on data stream clustering [15,3,13]. However, these algorithms are not well-suited to sensor networks due to their memory requirements and running time which depends on the number of data stream points.…”

“…We note that there is a large amount of prior work on data similarity and on time series similarity (see [15] for a survey of recent methods). However, previous work detects similarities using raw data.…”

“…One challenge associated with analyzing data streams is dealing with the large quantity of high-rate data. This has led to a number of summarization techniques (e.g., sampling, synopsis data structures, and aggregation [15,2]). However, these techniques are not wellsuited for sensor networks because of their computational complexity and memory requirements.…”

Time-series event-based prediction: An unsupervised learning framework based on genetic programming

“…In [13], the authors proposed an evolutionary approach based on GP to evolve mining rules from time-series. The proposed method is based on discretizing the timeseries using sliding window techniques to extract features to be divided into equal-size intervalsm and mapped into integer values to be classified into groups.…”

Symbolic Regression of Implicit Equations

“…Like clustering methods and other data mining approaches (McConaghy et al, 2008), unsupervised learning has the potential to find unexpected relationships in the data (De Falco et al, 2002;Mackin and Tazaki, 2000;Hetland and Saetrom, 2005). For example, unsupervised learning can create a model from positive examples only, and then use that model to detect outliers that do not belong to the original set.…”