Learning ECG Signal Features Without Backpropagation Using Linear Laws

TL;DR

This paper presents LLT-ECG, a physics-inspired, backpropagation-free method for ECG classification that automatically extracts features, achieving state-of-the-art results efficiently and verifiably.

Contribution

Introduces LLT-ECG, a novel physics-based approach for ECG feature extraction that operates without backpropagation, improving efficiency and interpretability.

Findings

Achieves state-of-the-art accuracy on PhysioNet ECG datasets.

Operates without backpropagation or hyperparameter tuning.

Provides a compact, verifiable feature representation.

Abstract

This paper introduces LLT-ECG, a novel method for electrocardiogram (ECG) signal classification that leverages concepts from theoretical physics to automatically generate features from time series data. Unlike traditional deep learning approaches, LLT-ECG operates in a forward manner, eliminating the need for backpropagation and hyperparameter tuning. By identifying linear laws that capture shared patterns within specific classes, the proposed method constructs a compact and verifiable representation, enhancing the effectiveness of downstream classifiers. We demonstrate LLT-ECG's state-of-the-art performance on real-world ECG datasets from PhysioNet, underscoring its potential for medical applications where speed and verifiability are crucial.