The emergence of powerful deep learning architectures has resulted in breakthrough innovations in several fields such as healthcare, precision farming, banking, education, and much more. Despite the advantages, there are limitations in deploying deep learning models in resource-constrained devices due to their huge memory size. This research work reports an innovative hybrid compression pipeline for compressing neural networks exploiting the untapped potential of z-score in weight pruning, followed by quantization using DBSCAN clustering and Huffman encoding. The proposed model has been experimented with state-of-the-art LeNet Deep Neural Network architectures using the standard MNIST and CIFAR datasets. Experimental results prove the compression performance of DeepCompNet by 26x without compromising the accuracy. The synergistic blend of the compression algorithms in the proposed model will ensure effortless deployment of neural networks leveraging DL applications in memory-constrained devices.

In medical informatics, the privacy, confidentiality and the integrity of patients’ data are an utmost concern, with legal and moral implications. So the patient information should be hidden in the medical images for ensuring confidentiality. This kind of operation needs reversible data hiding which extracts the hidden and cover data without any loss. In this paper, a novel reversible data hiding method is proposed for hiding data in medical image using edge detection and difference expansion technique. Edge detection is used to find Region of Interest (ROI). The Difference Expansion technique is used to hide data in both ROI and NROI regions. The novelty in this method is that it can hide more data in NROI region, resulting in high payload capacity and also guarantees the lossless recovery of both hidden data and medical image.