Book — 1 online resource (x, 173 pages) : illustrations Digital: text file.PDF.
Intro; Preface; Organization; Contents; Time Series Representation and Compression; Symbolic Representation of Time Series: A Hierarchical Coclustering Formalization; 1 Introduction; 2 Related Work; 3 Formalization of the SAXO Approach; 3.1 Prior Distribution of the SAXO Models; 3.2 Likelihood of Data Given a SAXO Model; 3.3 Evaluation Criterion; 4 Comparative Experiments on Real Datasets; 4.1 Coding Length Evaluation; 4.2 Supervised Learning Evaluation; 5 Conclusion and Perspectives; References; Dense Bag-of-Temporal-SIFT-Words for Time Series Classification; 1 Introduction; 2 Related Work
2.1 Distance-Based Time Series Classification2.2 Bag-of-Words for Time Series Classification; 2.3 Ensemble Classifiers for Time Series; 3 Bag-of-Temporal-SIFT-Words (BoTSW); 3.1 Keypoint Extraction in Time Series; 3.2 Description of the Extracted Keypoints; 3.3 Bag-of-Temporal-SIFT-Words for Time Series Classification; 4 Experiments and Results; 4.1 Experimental Setup; 4.2 Dense Extraction vs. Scale-Space Extrema Detection; 4.3 Impact of the BoW Normalization; 4.4 Comparison with State-of-the-Art Methods; 5 Conclusion; References
Dimension Reduction in Dissimilarity Spaces for Time Series Classification1 Introduction; 2 Related Work; 3 Dissimilarity Representations of Time Series; 3.1 The Basic Idea; 3.2 Dynamic Time Warping Distance; 3.3 Dissimilarity Representations; 3.4 Learning Classifiers in Dissimilarity Space; 3.5 Prototype Dependent Kernels; 4 Experiments; 4.1 Data; 4.2 Classifiers; 4.3 Experimental Protocol; 4.4 Results; 5 Conclusion; A Performance Profiles; References; Time Series Classification and Clustering; Fuzzy Clustering of Series Using Quantile Autocovariances; 1 Introduction
2 A Dissimilarity Based on Quantile Autocovariances3 Fuzzy Clustering Based on Quantile Autocovariances; 4 Simulation Study; 5 A Case Study; 6 Concluding Remarks; References; A Reservoir Computing Approach for Balance Assessment; 1 Introduction; 2 Balance Assessment Using Reservoir Computing; 3 Experimental Results; 4 Conclusions; References; Learning Structures in Earth Observation Data with Gaussian Processes; 1 Introduction; 2 Gaussian Process Regression; 2.1 Gaussian Processes: A Gentle Introduction; 2.2 On the Model Selection; 2.3 On the Covariance Function
2.4 Gaussian Processes Exemplified3 Advances in Gaussian Process Regression; 3.1 Structured, Non-stationary and Multiscale; 3.2 Time-based Covariance for GPR; 3.3 Heteroscedastic GPR: Learning the Noise Model; 3.4 Warped GPR: Learning the Output Transformation; 3.5 Source Code and Toolboxes; 4 Analysis of Gaussian Process Models; 4.1 Ranking Features Through the ARD Covariance; 4.2 Uncertainty Intervals; 5 Conclusions and Further Work; References; Monitoring Short Term Changes of Infectious Diseases in Uganda with Gaussian Processes; 1 Introduction; 2 Methods Used
This book constitutes the refereed proceedings of the First ECML PKDD Workshop, AALTD 2015, held in Porto, Portugal, in September 2016. The 11 full papers presented were carefully reviewed and selected from 22 submissions. The first part focuses on learning new representations and embeddings for time series classification, clustering or for dimensionality reduction. The second part presents approaches on classification and clustering with challenging applications on medicine or earth observation data. These works show different ways to consider temporal dependency in clustering or classification processes. The last part of the book is dedicated to metric learning and time series comparison, it addresses the problem of speeding-up the dynamic time warping or dealing with multi-modal and multi-scale metric learning for time series classification and clustering.