Learning representation with Wavelets an application to ECG classification

During this internship, under the supervision of Guillaume Metzler, we explored advanced methods to classify heartbeat patterns using electrocardiograms (ECGs) for arrhytmia detection. Our approach primarily hinged on Convolutional Neural Networks (CNNs) by imporving their capability to interpret heart signal data. Additionally, we delved into the Continuous Wavelet Transformation technique for signal analysis, providing a high-resolution insight into ECGs, with the expectation that it would give a richer and more appropriate description of the signal. In summary, our work represents an amalgamation of innovative techniques aimed at better representating and classifying heartbeat patterns using Deep Neural Networks and signal analysis transformations.

The report is available here (french).