Early classification on time series data has been found highly useful in a few important applications, such as medical and health informatics, industry production management, safety and security management. While some classifiers have been proposed to achieve good earliness in classification, the interpretability of early classification remains largely an open problem. Without interpretable features, application domain experts such as medical doctors may be reluctant to adopt early classification. In this paper, we tackle the problem of extracting interpretable features on time series for early classification. Specifically, we advocate local shapelets as features, which are segments of time series remaining in the same space of the input data and thus are highly interpretable. We extract local shapelets distinctly manifesting a target class locally and early so that they are effective for early classification. Our experimental results on seven benchmark real data sets clearly show that the local shapelets extracted by our methods are highly interpretable and can achieve effective early classification.

/pdf/extracting-interpretable-features-for-early-classification-c5yqqkexrq.pdf

Extracting Interpretable Features for Early Classification on Time Series.

In this paper, we formulate the problem of early classification of time series data, which is important in some time-sensitive applications such as health-informatics. We introduce a novel concept of MPL (Minimum Prediction Length) and develop ECTS (Early Classification on Time Series), an effective 1-nearest neighbor classification method. ECTS makes early predictions and at the same time retains the accuracy comparable to that of a 1NN classifier using the full-length time series. Our empirical study using benchmark time series data sets shows that ECTS works well on the real data sets where 1NN classification is effective.

/pdf/early-prediction-on-time-series-a-nearest-neighbor-approach-sv3s8i2a09.pdf

Early prediction on time series: a nearest neighbor approach

/pdf/time-series-knowledge-mining-4onltylg8y.pdf

Time Series Knowledge Mining

Supervised learning on sequence data, also known as sequence classification, has been well recognized as an important data mining task with many significant applications. Since temporal order is important in sequence data, in many critical applications of sequence classification such as medical diagnosis and disaster prediction, early prediction is a highly desirable feature of sequence classifiers. In early prediction, a sequence classifier should use a prefix of a sequence as short as possible to make a reasonably accurate prediction. To the best of our knowledge, early prediction on sequence data has not been studied systematically. In this paper, we identify the novel problem of mining sequence classifiers for early prediction. We analyze the problem and the challenges. As the first attempt to tackle the problem, we propose two interesting methods. The sequential classification rule (SCR) method mines a set of sequential classification rules as a classifier. A so-called early-prediction utility is defined and used to select features and rules. The generalized sequential decision tree (GSDT) method adopts a divide-and-conquer strategy to generate a classification model. We conduct an extensive empirical evaluation on several real data sets. Interestingly, our two methods achieve accuracy comparable to that of the stateof-the-art methods, but typically need to use only very short prefixes of the sequences. The results clearly indicate that early prediction is highly feasible and effective.

/pdf/mining-sequence-classifiers-for-early-prediction-2ah4up9d14.pdf

Mining Sequence Classifiers for Early Prediction

A new framework for analyzing sequential or temporal data such as time series is proposed. It differs from other approaches by the special emphasis on the interpretability of the results, since interpretability is of vital importance for knowledge discovery, that is, the development of new knowledge (in the head of a human) from a list of discovered patterns. While traditional approaches try to model and predict all time series observations, the focus in this work is on modelling local dependencies in multivariate time series. This makes it possible to deal with irregular or chaotic series. The proposed discovery process consists of (1) time series abstraction to get a representation close to the human perception of time series, (2) the enumeration and ranking of qualitative relationships in the data, (3) the specialization with quantitative constraints and the generalization of patterns to overcome limitations that are implicitly induced by the search bias.

Knowledge Discovery from Sequential Data

This work presents a system for supervised time series classification, capable of learning from series of different length and able of providing a classification when only part of the series are presented to the classifier. The induced classifiers consist of a linear combination of literals, obtained by boosting base classifiers that contain only one literal. Nevertheless, these literals are specifically designed for the task at hand and they test properties of fragments of the time series on temporal intervals. The method had already been developed for fixed length time series. This work exploits the symbolic nature of the classifier to add it two new features. First, the system has been slightly modified in order that it is now able to learn directly from variable length time series. Second, the classifier can be used to identify partial time series. This “early classification” is essential in some task, like on line supervision or diagnosis, where it is necessary to give an alarm signal as soon as possible. Several experiments on different data test are presented, which illustrate that the proposed method is highly competitive with previous approaches in terms of classification accuracy.

Boosting Interval-Based Literals: Variable Length and Early Classification

This work proposes a novel method for constructing RBF networks, based on boosting. The task assigned to the base learner is to select a RBF, while the boosting algorithm combines linearly the different RBFs. For each iteration of boosting a new neuron is incorporated into the network.

The method for selecting each RBF is based on randomly selecting several examples as the centers, considering the distances to these center as attributes of the examples and selecting the best split on one of these attributes. This selection of the best split is done in the same way than in the construction of decision trees. The RBF is computed from the center (attribute) and threshold selected.

This work is not about using RBFNs as base learners for boosting, but about constructing RBFNs by boosting.

Learning Classification RBF Networks by Boosting

Diagnosis of continuous dynamic systems: integrating consistency based diagnosis with machine-learning techniques

This work presents a learning system for the classification of multivariate time series This classification is useful in domains such as biomedical signals [9], continuous systems diagnosis [2] or data mining in temporal databases [3] 

Building RBF Networks for Time Series Classification by Boosting

Consistency-based diagnosis automatically provides fault detection and localization capabilities, using just models for correct behavior. However, it may exhibit a lack of discrimination power. Knowledge about fault modes can be added to tackle the problem. Unfortunately, it brings additional complexity issues, since it will be necessary to discriminate among a maximum of KN mode assignments, for N components and K possible fault modes per component. Usually, some kind of heuristic information is included in the diagnosis process to focus the model-based diagnostician. In this work we study the combination of a consistency-based diagnosis system together with a Case-based Reasoning system. The consistency-based diagnosis will perform fault detection and localization. The CBR system provides accurate indication of the most probable fault mode, at early stages of the localization process.

Juan José Rodríguez Diez

Papers

Boosting Interval-Based Literals: Variable Length and Early Classification

Learning Classification RBF Networks by Boosting

Diagnosis of continuous dynamic systems: integrating consistency based diagnosis with machine-learning techniques

Building RBF Networks for Time Series Classification by Boosting

Focusing fault localization in model-based diagnosis with case-based reasoning