analiza aiqbench
This research introduces ER-JEPA, a hierarchical self-supervised learning framework designed to leverage large unannotated ECG datasets for medical diagnosis tasks where labeled data is scarce. By integrating two Joint-Embedding Predictive Architectures with a Vision Transformer backbone, the model achieves state-of-the-art performance on multivariate time series benchmarks while maintaining low computational overhead. The approach demonstrates that structured representation learning can effectively extract local features and generalize across complex cardiac signal patterns without over-smoothing intermediate representations.
Hierarchical Self-Supervised Representation Learning Framework for Multivariate Time Series Grounded in ECG Analysis
arXiv:2607.01145v3 Announce Type: replace Abstract: Data analysis in the medical domain often encounters scenarios involving a limited target dataset and a large, unannotated dataset with a general distribution. Under such circumstances, self-supervised learning (SSL) methods are highly effective for utilizing large datasets, making them a popular choice for electrocardiogram (ECG) analysis. This work presents the Event Reconstruction Joint-Embedding Predictive Architecture (ER-JEPA), a lightweight SSL framework for multivariate time series, whose name and two-fold hierarchical structure are inspired by the diagnostic approach of cardiologists. At its core, ER-JEPA features: (1) a two-stage structure that constructs representations for each time interval and subsequently processes these representations as a univariate time series, (2) the hierarchical integration of two Joint-Embedding Predictive Architectures (JEPAs), and (3) a Vision Transformer (ViT) backbone. The structural concatenation of two JEPAs categorizes the model as a Hierarchical JEPA (H-JEPA), designed to encode multiple levels of abstract representations for enhanced prediction on complex tasks. This study reports a successful application of H-JEPA to 12-lead ECG data as a multivariate time series, alongside an analysis of the sensitivity of hierarchical representation during the pretraining stage. Furthermore, this study provides a qualitative demonstration that the intermediate representations produced by the first module of ER-JEPA excel at local feature extraction, as they are structurally free from over-smoothing. Pretrained on approximately 180,000 10-second recordings, the model achieves state-of-the-art downstream performance on the ST-MEM benchmark, with rapid computation and minimal resource usage.
Abstract:Data analysis in the medical domain often encounters scenarios involving a limited target dataset and a large, unannotated dataset with a general distribution. Under such circumstances, self-supervised learning (SSL) methods are highly effective for utilizing large datasets, making them a popular choice for electrocardiogram (ECG) analysis. This work presents the Event Reconstruction Joint-Embedding Predictive Architecture (ER-JEPA), a lightweight SSL framework for multivariate time series, whose name and two-fold hierarchical structure are inspired by the diagnostic approach of cardiologists. At its core, ER-JEPA features: (1) a two-stage structure that constructs representations for each time interval and subsequently processes these representations as a univariate time series, (2) the hierarchical integration of two Joint-Embedding Predictive Architectures (JEPAs), and (3) a Vision Transformer (ViT) backbone. The structural concatenation of two JEPAs categorizes the model as a Hierarchical JEPA (H-JEPA), designed to encode multiple levels of abstract representations for enhanced prediction on complex tasks. This study reports a successful application of H-JEPA to 12-lead ECG data as a multivariate time series, alongside an analysis of the sensitivity of hierarchical representation during the pretraining stage. Furthermore, this study provides a qualitative demonstration that the intermediate representations produced by the first module of ER-JEPA excel at local feature extraction, as they are structurally free from over-smoothing. Pretrained on approximately 180,000 10-second recordings, the model achieves state-of-the-art downstream performance on the ST-MEM benchmark, with rapid computation and minimal resource usage.
29 pages, 8 figures. Further clarified, added the new downstream task in the abstract and intro since last update.<< Polished text, improved formatting, fixed speed benchmark result, and added new downstream task. Code will be made publicly available soon
Machine Learning (cs.LG); Signal Processing (eess.SP)
arXiv:2607.01145 [cs.LG]
(or arXiv:2607.01145v3 [cs.LG] for this version)
Text integral sub licența deschisă a publicației · 285 cuvinte
Subiecte conexe
CF-JEPA: Mask-free forward prediction with asymmetric encoder utilization for time-series representation learning
arXiv · 8 iun. 2026
RECTOR: Masked Region-Channel-Temporal Modeling for Affective and Cognitive Representation Learning
arXiv · 16 iun. 2026
STAR: Adaptive Spatial-Temporal Normalization for Unified Microservice Incident Management
arXiv · 30 sept. 2026
NV-Tesseract-AD: Curriculum-Enhanced Conditional Diffusion for Multivariate Time-Series Anomaly Detection
arXiv · 12 oct. 2026