InceptionTime: Finding AlexNet for Time Series Classification

Abstract

This paper brings deep learning at the forefront of research into Time Series Classification (TSC). TSC is the area of machine learning tasked with the categorization (or labelling) of time series. The last few decades of work in this area have led to significant progress in the accuracy of classifiers, with the state of the art now represented by the HIVE-COTE algorithm. While extremely accurate, HIVE-COTE cannot be applied to many real-world datasets because of its high training time complexity in O(N2 * T4) for a dataset with N time series of length T. For example, it takes HIVE-COTE more than 8 days to learn from a small dataset with N = 1500 time series of short length T = 46. Meanwhile deep learning has received enormous attention because of its high accuracy and scalability. Recent approaches to deep learning for TSC have been scalable, but less accurate than HIVE-COTE. We introduce InceptionTime - an ensemble of deep Convolutional Neural Network (CNN) models, inspired by the Inception-v4 architecture. Our experiments show that InceptionTime is on par with HIVE-COTE in terms of accuracy while being much more scalable: not only can it learn from 1,500 time series in one hour but it can also learn from 8M time series in 13 hours, a quantity of data that is fully out of reach of HIVE-COTE.

References in corpus (7)

• Time Warp Edit Distance with Stiffness Adjustment for Time Series Matching
• Fast and Accurate Time Series Classification with WEASEL
• Transfer learning for time series classification
• Proximity Forest: An effective and scalable distance-based classifier for time series
• Adversarial Attacks on Deep Neural Networks for Time Series Classification
• Deep Neural Network Ensembles for Time Series Classification
• Automatic alignment of surgical videos using kinematic data

Cited by in corpus (92)

• ROCKET: Exceptionally fast and accurate time series classification using random convolutional kernels
• Benchmarking Multivariate Time Series Classification Algorithms
• MINIROCKET: A Very Fast (Almost) Deterministic Transform for Time Series Classification
• Self-attention for raw optical Satellite Time Series Classification
• HIVE-COTE 2.0: a new meta ensemble for time series classification
• Improving Position Encoding of Transformers for Multivariate Time Series Classification
• Bake off redux: a review and experimental evaluation of recent time series classification algorithms
• Satellite Image Time Series Analysis for Big Earth Observation Data
• TodyNet: Temporal Dynamic Graph Neural Network for Multivariate Time Series Classification
• Time-Series Classification in Smart Manufacturing Systems: An Experimental Evaluation of State-of-the-Art Machine Learning Algorithms
• A Spatiotemporal Deep Neural Network for Fine-Grained Multi-Horizon Wind Prediction
• TCGAN: Convolutional Generative Adversarial Network for Time Series Classification and Clustering
• Decoding EEG Brain Activity for Multi-Modal Natural Language Processing
• Uncertainty-Aware Deep Ensembles for Reliable and Explainable Predictions of Clinical Time Series
• Label-efficient Time Series Representation Learning: A Review
• The FreshPRINCE: A Simple Transformation Based Pipeline Time Series Classifier
• Domain Adaptation for Time-Series Classification to Mitigate Covariate Shift
• TimeMAE: Self-Supervised Representations of Time Series with Decoupled Masked Autoencoders
• A comparative study of different machine learning methods for dissipative quantum dynamics
• Random Dilated Shapelet Transform: A New Approach for Time Series Shapelets
• Autoregressive Coefficients based Intelligent Protection of Transmission Lines Connected to Type-3 Wind Farms
• DeepVATS: Deep Visual Analytics for Time Series
• An Evaluation of Machine Learning and Deep Learning Models for Drought Prediction using Weather Data
• Long-Short Ensemble Network for Bipolar Manic-Euthymic State Recognition Based on Wrist-worn Sensors
• Privacy Meets Explainability: A Comprehensive Impact Benchmark
• Multi-modal Heart Failure Risk Estimation based on Short ECG and Sampled Long-Term HRV
• A Systematic Literature Review of Spatio-Temporal Graph Neural Network Models for Time Series Forecasting and Classification
• Unsupervised Feature Based Algorithms for Time Series Extrinsic Regression
• Explainable Machine Learning for Breakdown Prediction in High Gradient RF Cavities
• Appliance Detection Using Very Low-Frequency Smart Meter Time Series
• Detecting COVID-19 from digitized ECG printouts using 1D convolutional neural networks
• Auxiliary Cross-Modal Representation Learning with Triplet Loss Functions for Online Handwriting Recognition
• TELESTO: A Graph Neural Network Model for Anomaly Classification in Cloud Services
• SEAnet: A Deep Learning Architecture for Data Series Similarity Search
• NILMFormer: Non-Intrusive Load Monitoring that Accounts for Non-Stationarity
• Evaluation of (Un-)Supervised Machine Learning Methods for GNSS Interference Classification with Real-World Data Discrepancies
• Ball Trajectory Inference from Multi-Agent Sports Contexts Using Set Transformer and Hierarchical Bi-LSTM
• Deep learning in automated ultrasonic NDE -- developments, axioms and opportunities
• Convolutional and Deep Learning based techniques for Time Series Ordinal Classification
• Neural network time-series classifiers for gravitational-wave searches in single-detector periods
• QCore: Data-Efficient, On-Device Continual Calibration for Quantized Models -- Extended Version
• Generalizable deep learning for photoplethysmography-based blood pressure estimation -- A Benchmarking Study
• Enhancing Eye-Tracking Performance through Multi-Task Learning Transformer
• STADEE: STAtistics-based DEEp Detection of Machine Generated Text
• Exploring Kolmogorov-Arnold Networks for Interpretable Time Series Classification
• Quantifying Quality of Class-Conditional Generative Models in Time-Series Domain
• A General Framework for Generative Self-supervised Learning in Non-invasive Estimation of Physiological Parameters Using Photoplethysmography
• Spikebench: An open benchmark for spike train time-series classification
• Automatic Feature Engineering for Time Series Classification: Evaluation and Discussion
• Deep Semi-Supervised Learning for Time Series Classification
• Towards Data-Driven Electricity Management: Multi-Region Harmonized Data and Knowledge Graph
• Estimating Motor Symptom Presence and Severity in Parkinson's Disease from Wrist Accelerometer Time Series using ROCKET and InceptionTime
• Combining Multi-Objective Bayesian Optimization with Reinforcement Learning for TinyML
• Recurrence Plot and Change Quantile-based Deep Supervised and Semi-supervised Protection for Transmission Lines Connected to Photovoltaic Plants
• Diagnostic Spatio-temporal Transformer with Faithful Encoding
• Leveraging Deep Learning for Time Series Extrinsic Regression in predicting photometric metallicity of Fundamental-mode RR Lyrae Stars
• ADF & TransApp: A Transformer-Based Framework for Appliance Detection Using Smart Meter Consumption Series
• Reducing Label Dependency in Human Activity Recognition with Wearables: From Supervised Learning to Novel Weakly Self-Supervised Approaches
• Comparison of neural network architectures for feature extraction from binary black hole merger waveforms
• Time Distributed Deep Learning Models for Purely Exogenous Forecasting: Application to Water Table Depth Prediction using Weather Image Time Series
• Season-Independent PV Disaggregation Using Multi-Scale Net Load Temporal Feature Extraction and Weather Factor Fusion
• Intelligent Pooling: Proactive Resource Provisioning in Large-scale Cloud Service
• Interpretable Feature Construction for Time Series Extrinsic Regression
• Time delay estimation in unresolved lensed quasars
• Towards Diverse Perspective Learning with Selection over Multiple Temporal Poolings
• Deep Unsupervised Domain Adaptation for Time Series Classification: a Benchmark
• DeepHeteroIoT: Deep Local and Global Learning over Heterogeneous IoT Sensor Data
• From Private to Public: Benchmarking GANs in the Context of Private Time Series Classification
• An Empirical Evaluation of Multivariate Time Series Classification with Input Transformation across Different Dimensions
• Fingerprint Recognition of Partial Discharge Signals in Deep Learning Enhanced Rydberg Atomic Sensors
• ATCN: Resource-Efficient Processing of Time Series on Edge
• Attention-Based Fusion of IQ and FFT Spectrograms with AoA Features for GNSS Jammer Localization
• Behavioral Biometrics for Automatic Detection of User Familiarity in VR
• Conspiracy to Commit: Information Pollution, Artificial Intelligence, and Real-World Hate Crime
• RED CoMETS: An ensemble classifier for symbolically represented multivariate time series
• Concept Extraction for Time Series with ECLAD-ts
• Class-Dependent Perturbation Effects in Evaluating Time Series Attributions
• ExChanGeAI: An End-to-End Platform and Efficient Foundation Model for Electrocardiogram Analysis and Fine-tuning
• A 1-D CNN inference engine for constrained platforms
• Few Labels are all you need: A Weakly Supervised Framework for Appliance Localization in Smart-Meter Series
• NeuroSleepNet: An Explainable Multi-Head Attention-Based Framework With Spatial and Multi-Scale Independent Temporal Context Learning for Automatic Sleep Stage Scoring
• AI-driven Orchestration at Scale: Estimating Service Metrics on National-Wide Testbeds
• Deep Learning for Time Series Classification of Parkinson's Disease Eye Tracking Data
• Binary Classification of Light and Dark Time Traces of a Transition Edge Sensor Using Convolutional Neural Networks
• HIT-ROCKET: Hadamard-vector Inner-product Transformer for ROCKET
• Enhancing Time Series Classification with Diversity-Driven Neural Network Ensembles
• A Standardized Benchmark for Skeleton-Based Rehabilitation Assessment Using Deep Learning
• Enhancing deep learning models for time series classification via knowledge distillation
• Boosting Certified Robustness for Time Series Classification with Efficient Self-Ensemble
• MIHT: A Hoeffding Tree for Time Series Classification using Multiple Instance Learning
• FREQuency ATTribution: benchmarking frequency-based occlusion for time series data
• Using matrix-product states for time-series machine learning