All projects
OngoingApplied AI/MLApplied AI/ML · Signal Processing · TDA
Topological Anomaly Detection in ECG Signals
Exploring topological data analysis for detecting anomalous ECG signals, and comparing it against conventional signal features.
Objective
Explore topological data analysis for detecting anomalous ECG signals.
Methods planned
- ECG preprocessing.
- Beat segmentation.
- Time-delay embedding.
- Persistent homology.
- Persistence images.
- Persistence landscapes.
- Betti curves and entropy.
- Isolation Forest.
- One-Class SVM.
- Comparison with conventional signal features.
Data sources
- MIT-BIH Arrhythmia Database (intended real dataset)