Shallow Learning vs. Deep Learning in Image Processing

This chapter compares shallow learning and deep learning techniques for image processing applications. Shallow learning (SL) algorithms use specific features that are defined from the input data for the model and have one or two layered models. Deep learning (DL) eliminates some of the data pre-processing that is typically involved with shallow learning. These algorithms can ingest and process unstructured data, like text and images, and it automates feature extraction, removing some of the dependency on human experts. DL and SL show enormous growth in the medical field. These algorithms are important to predict the symptoms of early disease. ECG graphs are the main diagnostic tool for cardiac diseases. As heartbeats are a vital indicator of heart disease, their classification is essential. In this chapter SL and DL methods have been compared using ECG heartbeat images, and it is shown that although the DL method has some advantages, the SL method also can be applied and achieves a high accuracy rate for image classification.