Distributed and parallel time series feature extraction for industrial big data applications
arXiv:1610.07717
Abstract
The all-relevant problem of feature selection is the identification of all strongly and weakly relevant attributes. This problem is especially hard to solve for time series classification and regression in industrial applications such as predictive maintenance or production line optimization, for which each label or regression target is associated with several time series and meta-information simultaneously. Here, we are proposing an efficient, scalable feature extraction algorithm for time series, which filters the available features in an early stage of the machine learning pipeline with respect to their significance for the classification or regression task, while controlling the expected percentage of selected but irrelevant features. The proposed algorithm combines established feature extraction methods with a feature importance filter. It has a low computational complexity, allows to start on a problem with only limited domain knowledge available, can be trivially parallelized, is highly scalable and based on well studied non-parametric hypothesis tests. We benchmark our proposed algorithm on all binary classification problems of the UCR time series classification archive as well as time series from a production line optimization project and simulated stochastic processes with underlying qualitative change of dynamics.
References in corpus (2)
Cited by in corpus (28)
- Parallel Spatio-Temporal Attention-Based TCN for Multivariate Time Series Prediction
- Machine learning search for variable stars
- Learning Unit State Recognition Based on Multi-channel Data Fusion
- Curvature-based Feature Selection with Application in Classifying Electronic Health Records
- Feature engineering workflow for activity recognition from synchronized inertial measurement units
- Comparison of gait phase detection using traditional machine learning and deep learning techniques
- Not to Cry Wolf: Distantly Supervised Multitask Learning in Critical Care
- Clumpiness: Time-domain classification of red-giant evolutionary states
- Adversarial Framework with Certified Robustness for Time-Series Domain via Statistical Features
- Automated Machine Learning for Remaining Useful Life Predictions
- Machine learning-based classification for Single Photon Space Debris Light Curves
- Activity Archetypes in Question-and-Answer (Q&A) Websites - A Study of 50 Stack Exchange Instances
- Automatic Feature Engineering for Time Series Classification: Evaluation and Discussion
- Towards a Near Universal Time Series Data Mining Tool: Introducing the Matrix Profile
- Variability search in M 31 using Principal Component Analysis and the Hubble Source Catalog
- Identification of Internal Faults in Indirect Symmetrical Phase Shift Transformers Using Ensemble Learning
- Can we Estimate Truck Accident Risk from Telemetric Data using Machine Learning?
- Forecasting with sktime: Designing sktime's New Forecasting API and Applying It to Replicate and Extend the M4 Study
- KPIs-Based Clustering and Visualization of HPC jobs: a Feature Reduction Approach
- Modelling of Networked Measuring Systems -- From White-Box Models to Data Based Approaches
- A Novel Hybrid Feature Importance and Feature Interaction Detection Framework for Predictive Optimization in Industry 4.0 Applications
- Multi-Statistic Approximate Bayesian Computation with Multi-Armed Bandits
- The ETS challenges: a machine learning approach to the evaluation of simulated financial time series for improving generation processes
- Probabilistic Modeling for Novelty Detection with Applications to Fraud Identification
- Distance Rank Score: Unsupervised filter method for feature selection on imbalanced dataset
- Local Exceptionality Detection in Time Series Using Subgroup Discovery
- Unobtrusive and Multimodal Approach for Behavioral Engagement Detection of Students
- A Survey on Data Processing Methods and Cloud Computation