MINIROCKET: A Very Fast (Almost) Deterministic Transform for Time Series Classification

Abstract

Until recently, the most accurate methods for time series classification were limited by high computational complexity. ROCKET achieves state-of-the-art accuracy with a fraction of the computational expense of most existing methods by transforming input time series using random convolutional kernels, and using the transformed features to train a linear classifier. We reformulate ROCKET into a new method, MINIROCKET, making it up to 75 times faster on larger datasets, and making it almost deterministic (and optionally, with additional computational expense, fully deterministic), while maintaining essentially the same accuracy. Using this method, it is possible to train and test a classifier on all of 109 datasets from the UCR archive to state-of-the-art accuracy in less than 10 minutes. MINIROCKET is significantly faster than any other method of comparable accuracy (including ROCKET), and significantly more accurate than any other method of even roughly-similar computational expense. As such, we suggest that MINIROCKET should now be considered and used as the default variant of ROCKET.

arXiv comment10 pages, 11 figures; Updated to accepted version

References in corpus (7)

• Deep learning for time series classification: a review
• InceptionTime: Finding AlexNet for Time Series Classification
• Quantized Neural Networks: Training Neural Networks with Low Precision Weights and Activations
• ROCKET: Exceptionally fast and accurate time series classification using random convolutional kernels
• Proximity Forest: An effective and scalable distance-based classifier for time series
• Should we really use post-hoc tests based on mean-ranks?
• A tale of two toolkits, report the third: on the usage and performance of HIVE-COTE v1.0

Cited by in corpus (20)

• Improving Position Encoding of Transformers for Multivariate Time Series Classification
• Bake off redux: a review and experimental evaluation of recent time series classification algorithms
• An ensemble of convolution-based methods for fault detection using vibration signals
• Appliance Detection Using Very Low-Frequency Smart Meter Time Series
• RandomNet: Clustering Time Series Using Untrained Deep Neural Networks
• Automatic Feature Engineering for Time Series Classification: Evaluation and Discussion
• FedST: Secure Federated Shapelet Transformation for Time Series Classification
• Estimating Motor Symptom Presence and Severity in Parkinson's Disease from Wrist Accelerometer Time Series using ROCKET and InceptionTime
• NFTracer: Tracing NFT Impact Dynamics in Transaction-flow Substitutive Systems with Visual Analytics
• COSCO: A Sharpness-Aware Training Framework for Few-shot Multivariate Time Series Classification
• Extracting, Visualizing, and Learning from Dynamic Data: Perfusion in Surgical Video for Tissue Characterization
• ATCN: Resource-Efficient Processing of Time Series on Edge
• Motor Imagery EEG Signal Classification Using Minimally Random Convolutional Kernel Transform and Hybrid Deep Learning
• Revisit Time Series Classification Benchmark: The Impact of Temporal Information for Classification
• T-Rep: Representation Learning for Time Series using Time-Embeddings
• Rock the KASBA: Blazingly Fast and Accurate Time Series Clustering
• MSAD: A Deep Dive into Model Selection for Time series Anomaly Detection
• Explainable classification of astronomical uncertain time series
• Few Labels are all you need: A Weakly Supervised Framework for Appliance Localization in Smart-Meter Series
• The largest ground-based catalogue of M-dwarf flares