“…Among the most prevalent types of neural networks, convolutional neural networks (CNNs) were used by Ahuja et al [12]. Proenca et al have attempted to improve the feature learning process using CNNs by implicitly identifying the areas of interest in the input data that should be prioritized, rather than blocking off any areas in the test/training samples [13]. For supervised learning, they used a four-layer stacked convolutional network followed by a 512-dimensional feature vector, along with cosine similarity for testing [14].…”