The UCR Time Series Archive expands from 85 to 128 datasets, offering advice and insights.
problem Lack of comprehensive data sets for time series analysis.
method Periodic expansions of the archive, providing advice and novel insights.
result A significant increase in the number of datasets from 85 to 128.
Temporal information impacts only a fraction of time series datasets, skewing benchmark evaluations.
problem Temporal information's impact on time series classification is often overestimated.
method Permutation tests on UCR archive to identify datasets where temporal info is irrelevant.
result Many tabular datasets perform well without temporal info, skewing benchmark evaluations.
Proximity Forest classifies time series in milliseconds from large datasets.
problem Classifying time series from large datasets with high accuracy and speed.
method Ensemble of randomized Proximity Trees, leveraging proximity measures instead of attribute values.
result Proximity Forest achieves high accuracy on large datasets and is significantly faster than state-of-the-art models.
Benchmark study evaluates 8 clustering methods on 99 UCR time series datasets.
problem Assessing the performance of clustering methods on time series data.
method Examines 8 clustering methods across 3 categories and 3 distance measures on 99 UCR datasets.
result Provides a comprehensive dataset-level assessment of clustering methods.
New MTSC archive adds 30 multivariate time series datasets for evaluation.
problem Lack of multivariate time series datasets for rigorous evaluation.
method Forming the first MTSC archive with 30 datasets, equal lengths, no missing data, and train/test splits.
result Addresses the need for a comprehensive evaluation of multivariate time series classification algorithms.
Recent anomaly detection benchmarks are flawed, potentially misleading progress.
problem Flawed benchmark datasets create misleading progress reports.
method Identified four flaws in benchmark datasets and introduced a new archive.
result Published comparisons may be unreliable due to flaws in benchmark datasets.
Introduces a new benchmark for time series extrinsic regression.
problem Predicting a single continuous value from univariate or multivariate time series, not necessarily related to the predictor.
method Developed a new benchmarking archive for time series extrinsic regression.
result Initial benchmarking of existing models on the new TSER datasets.
Adaptive weighting schemes enhance time-series data augmentation for financial and UCR datasets.
problem Limited size of time-series datasets hinders model performance.
method Two adaptive weighting schemes for automatic data augmentation.
result Improves annualized returns by over 50% on financial dataset and outperforms state-of-the-art on half of UCR datasets.
Deep neural networks outperform traditional ensemble methods in time series classification.
problem Deep learning models struggle to match traditional ensemble methods in time series classification.
method Developed an ensemble of 60 deep learning models to improve time series classification performance.
result The proposed Neural Network Ensemble (NNE) outperforms current state-of-the-art methods.
New study on time series anomaly detection shows overlapping inference improves performance.
problem Heterogeneous evaluation practices and inference procedures in time series anomaly detection.
method Unified training, tuning, and evaluation protocol on TSB-AD benchmark, analyzing overlapping vs. disjoint inference.
result Overlapping inference yields consistent improvements, with average relative gain up to +28%.
A novel time series clustering method that considers segment typologies.
problem Lack of consideration for the similarity of different subsequences in time series clustering.
method Two-stage clustering: polynomial segmentation followed by hierarchical clustering of segments, then final clustering of time series.
result The method outperforms state-of-the-art techniques on UCR Time Series Classification Archive datasets.
A new method extracts features from time series data using iterated sums and improves classification accuracy.
problem Time series classification challenges.
method Feature extraction using iterated-sums signature (ISS) followed by a linear classifier.
result Competitive with state-of-the-art methods on UCR archive.
Paper proposes a robust time series classification method using ResNet and Recurrence Plots.
problem Classifying time series data is challenging and underexplored.
method Transfer learning in Deep Neural Networks, 2D Recurrence Plots, ResNet architecture, simplified preprocessing.
result First time multi-time series classification using a single network.
Transfer learning improves TSC models by 71 out of 85 datasets.
problem Improving deep CNNs for TSC using transfer learning.
method Pre-trained deep CNNs fine-tuned on various UCR datasets.
result Transfer learning can improve or degrade TSC model predictions.
RED CoMETS improves multivariate time series classification accuracy.
problem Complexity of multivariate time series classification.
method Ensemble classifier RED CoMETS for symbolically represented multivariate time series.
result RED CoMETS achieves highest reported accuracy on 'HandMovementDirection' dataset.
Computer vision SSL methods show effectiveness on time series data.
problem Evaluate if computer vision SSL frameworks are effective on time series data.
method Evaluated on UCR and UEA archives, proposed a new method improving VICReg.
result Computer vision SSL frameworks can be effective on time series data.
TimeVQVAE-AD detects anomalies in time series data with high accuracy and provides explainable results.
problem Detecting and explaining anomalies in time series data accurately.
method Masked latent generative modeling in time-frequency domain.
result TimeVQVAE-AD outperforms existing methods in anomaly detection and explainability.
MINIROCKET speeds up time series classification by 75x.
problem High computational complexity in time series classification.
method Random convolutional kernels and linear classifier.
result Up to 75 times faster on larger datasets.
Guided warping augments time series data by aligning features with a teacher.
problem Small time series datasets limit neural network performance.
method Guided warping with a discriminative teacher to augment data deterministically.
result Significant improvement in performance on various time series datasets.
Archive of 20 time series datasets for forecasting evaluation.
problem Lack of comprehensive time series forecasting datasets.
method Compilation and characterisation of 20 datasets from various domains.
result Characterisation and performance evaluation of datasets.
RST improves environmental time series classification accuracy using randomized B-spline trees.
problem Improving accuracy in classifying complex environmental time series.
method Randomized Spline Trees (RST) integrates randomized functional representations into ensemble learning.
result RST variants outperform standard Random Forests and Gradient Boosting on most environmental time series datasets.
TimeVQVAE uses VQ for better time series generation.
problem Training GANs and RNNs for time series generation have limitations.
method Vector quantization with bidirectional transformer priors in time-frequency domains.
result Generates high-quality synthetic signals with better temporal consistency.
Paper proposes DTW-SOM for visual exploration of time-series motifs.
problem Exploring motifs extracted by time-series motif discovery algorithms.
method Adapted Self-Organizing Map (DTW-SOM) using Dynamic Time Warping distance.
result DTW-SOM effectively visualizes motifs from time-series data.
NM-VQTSG improves synthetic time series fidelity by aligning distributions.
problem Fidelity challenges in VQ-based time series generation.
method Neural mapping model using U-Net to refine synthetic data.
result Significant improvements in FID, IS, and conditional FID metrics.
End-to-end model for time series classification with missing data.
problem Time series classification with missing data.
method End-to-end neural network that unifies imputation and representation learning.
result Proposed model outperforms state-of-the-art approaches for incomplete time series classification.
Deep learning improves time series classification.
problem Challenging task of classifying time series data.
method Empirical study of deep learning architectures for TSC.
result Deep learning achieves state-of-the-art performance in TSC.
LETS-GZSL tackles GZSL for time series classification, achieving high accuracy.
problem Recognizing unseen classes from time series data when only seen examples are labeled.
method Embedding-based approach combined with attribute vectors.
result Achieves a harmonic mean of at least 55% on most UCR datasets.
Improved scalability of BOSS ensemble for time series classification.
problem Non-trivial scalability issues in BOSS and WEASEL.
method Random selection of classifiers, ensembling techniques, and build time reduction.
result Significant reduction in build time with minimal accuracy loss.
Enhanced Sampling Scheme improves masked generative modeling.
problem Limitations of existing sampling schemes in masked non-autoregressive generative modeling.
method ESS consists of three stages: Naive Iterative Decoding, Critical Reverse Sampling, and Critical Resampling.
result ESS achieves significant performance gains in unconditional and class-conditional sampling.
TS-CHIEF improves TSC accuracy while drastically reducing runtime.
problem Infeasibility of HIVE-COTE for large datasets.
method Combines time series embeddings into a tree-structured classifier.
result Achieves state-of-the-art accuracy with 130k time series in 2 days.
Co-eye combines multiple symbolic representations to improve time series classification accuracy.
problem Challenges in time series classification due to domain diversity.
method Inspired by compound eyes, Co-eye uses multiple symbolic representations and hyper-parameterised lenses to classify time series data.
result Co-eye outperforms state-of-the-art techniques in accuracy and robustness across various domains.
Study compares time series classification algorithms using simulated data.
problem Understanding why some TSC algorithms outperform others.
method Design and implement simulators for five feature spaces, observe classifier performance.
result Ensemble methods often outperform single algorithms, especially when data representation is unknown.
Meta-learning improves few-shot time series classification.
problem Few labeled data for time series classification.
method Gradient-based meta-learning for residual neural networks.
result Meta-learning outperforms baselines on 41 datasets.
The paper critiques current time series classification evaluation methods.
problem Current performance evaluation methods in time series classification are criticized.
method No specific new method proposed, but a critical analysis of existing methods.
result Suggests a need for discussion and reflection on TSC performance evaluation.
New metrics fail adversarial tests, with some more robust than others.
problem Evaluation metrics for time-series anomaly detection were improved but not fully robust.
method Adversarial stress-testing of 12 adopted metrics on real benchmarks.
result Some metrics are more robust than others, with ROC-based metrics being gamed more often.
New TSER algorithms outperform existing methods in time series extrinsic regression.
problem Improving time series extrinsic regression models.
method Extended TSER archive, introduced two new algorithms (FreshPRINCE and DrCIF), compared with rotation forest.
result DrCIF and FreshPRINCE models significantly outperform existing methods.
New algorithms benchmarked for multivariate time series classification.
problem Comparing algorithms for multivariate time series classification.
method Review and comparison of recent MTSC algorithms using the UEA archive.
result HIVE-COTE ensemble is most accurate for MTSC, but dynamic time warping is competitive.
Fusion framework improves time series classification across different datasets.
problem Kernel-based methods like Rocket perform inconsistently across datasets.
method Fusion-3 framework that adaptively fuses three representations (Rocket, SAX, SFA) based on dataset properties.
result Fusion-3 framework yields small but consistent average improvements over Rocket on 113 UCR datasets.
TTW aligns time-series faster and more accurately than existing methods.
problem Efficiently aligning multiple time-series signals with varying lengths.
method TTW uses a sinc convolutional kernel and gradient-based optimization for linear time and sequence complexity.
result TTW outperforms existing methods in time-series averaging and classification tasks.
MDF represents time series motifs as images for improved classification.
problem Classifying time series data with high-order patterns.
method Motif Difference Field (MDF) using Fully Convolutional Networks (FCN).
result MDF outperforms other methods on UCR time series datasets.
UniShape improves time series classification by selecting relevant subsequences.
problem Classifying time series data requires capturing interpretable shapelets.
method UniShape uses a shape-aware adapter to aggregate multiscale subsequences into class tokens.
result UniShape achieves state-of-the-art classification performance.
Paper proposes ATN to attack time series classification models.
problem Adversarial attacks on time series classification models.
method Adversarial transformation network (ATN) on a distilled model.
result Time series classification models are susceptible to adversarial attacks.
Unified pipeline classifies time series using complex networks and persistent homology.
problem Classifying univariate time series using various graph constructions and metrics.
method Time series to graph, graph to dissimilarity matrix, filtration to persistence diagrams, vectorization to features.
result Persistence-based features are robust to noise and optimal graph type depends on signal structure.
Paper proposes transforming ATN to attack multivariate time series models.
problem Generating adversarial samples for multivariate time series classification models.
method Proposes using a distilled model as a surrogate to mimic attacked models and applies 1-NN DTW and FCN attacks.
result Both models were susceptible to attacks on all 18 datasets.
Paper studies Time Series Extrinsic Regression, improving on existing methods.
problem Learning relationship between time series and continuous scalar variables.
method Motivated and studied TSER, benchmarked TSC and ML algorithms.
result Rocket algorithm achieves highest overall accuracy in TSER.
Automated smoothing does not significantly improve time series classification performance.
problem Improving time series classification algorithms using automated smoothing methods.
method Assessed six smoothing algorithms (moving average, exponential, etc.) on three benchmark classifiers.
result No significant improvement in performance over unsmoothed data.
Proposes mWDN for interpreting time series analysis.
problem Lack of effective modeling for frequency information in time series analysis.
method Wavelet-based multilevel neural network structure (mWDN).
result Demonstrates excellent performance and interpretability of mWDN models.
VOLARE provides standardized realized volatility measures from financial data.
problem Lack of standardized realized volatility measures from ultra-high-frequency data.
method Asset-specific pipeline for cleaning and sampling data, providing a wide range of realized estimators.
result Comprehensive set of realized estimators for equities, exchange rates, and futures.