Intro
Preface
Workshop Description
Organization
Causal Discovery in Observational Time Series (Invited Talk)
Contents
Oral Presentation
Adjustable Context-Aware Transformer
1 Introduction
2 Problem Definition
3 Related Work
4 Methodology
4.1 Background: Issues Arising from Point-Wise Attention
4.2 Temporal Attention
4.3 Adjustable Context-Aware Attention
4.4 Efficient Adjustable Context-Aware Attention
4.5 Overarching Architecture
5 Experiments
5.1 Datasets
5.2 Evaluation Metrics
5.3 Baselines

5.4 Model Training and Hyperprarameters
5.5 Results and Discussion
6 Conclusion
References
Clustering of Time Series Based on Forecasting Performance of Global Models
1 Introduction
2 A Clustering Algorithm Based on Prediction Accuracy of Global Forecasting Models
3 Simulation Study
3.1 Experimental Design
3.2 Alternative Approaches and Assessment Criteria
3.3 Results and Discussion
4 Application to Real Data
5 Conclusions
References
Experimental Study of Time Series Forecasting Methods for Groundwater Level Prediction
1 Introduction

2 Data Collection
3 Groundwater Level Forecasting
3.1 Local Versus Global Time Series Forecasting
3.2 Considered Methods
4 Experimental Settings
4.1 Setup
4.2 Comparison Metrics
5 Results
5.1 Generalized Autoregressive Models Results
5.2 DeepAR-Based Models Results
5.3 Prophet-Based Models Results
5.4 Comparing the Three Groups of Models
5.5 Discussion
6 Conclusion
References
Fast Time Series Classification with Random Symbolic Subsequences
1 Introduction
2 Related Work
3 Proposed Method
3.1 Symbolic Representations of Time Series

3.2 MrSQM Variants
4 Evaluation
4.1 Experiment Setup
4.2 Sensitivity Analysis
4.3 MrSQM Versus State-of-the-Art Symbolic Time Series Classifiers
4.4 MrSQM Versus Other State-of-the-Art Time Series Classifiers
5 Conclusion
References
RESIST: Robust Transformer for Unsupervised Time Series Anomaly Detection
1 Introduction
2 Related Work
3 Method
3.1 RESIST Architecture
3.2 Robust Training Loss
3.3 Hypotheses
4 Experiments and Results
4.1 Dataset Description
4.2 Data Preprocessing
4.3 Training and Testing Protocols

4.4 Training Parameter Settings and Evaluation Criteria
4.5 Results
5 Conclusion and Perspectives
References
Window Size Selection in Unsupervised Time Series Analytics: A Review and Benchmark
1 Introduction
2 Background and Related Work
2.1 Definitions
2.2 Anomaly Detection
2.3 Segmentation
2.4 Motif Discovery
3 Window Size Selection
3.1 Dominant Fourier Frequency
3.2 Highest Autocorrelation
3.3 Hybrids: AutoPeriod and RobustPeriod
3.4 Multi-Window-Finder
3.5 Summary Statistics Subsequence
4 Experimental Evaluation
4.1 Setup