An Ensemble Approach to Improve Homomorphic Encrypted Data Classification Performance

Dana Alsagheer and Hadi Mansourifar, University of Houston

ABSTRACT

Homomorphic encryption (HE) permits users to perform computations on encrypted data without first decrypting it. HE can be used for privacy-preserving outsourced computation and analysis, allowing data to be encrypted and outsourced to commercial cloud environments for processing while encrypted or sensitive data. HE enables new services by removing privacy barriers inhibiting data sharing or increasing the security of existing services. A convolution neural network (CNN) can be homomorphically evaluated using addition and multiplication by replacing the activation function, such as Rectified Linear Units (ReLU), with a low polynomial degree. To achieve the same performance as the ReLU activation function, we study the impact of applying the ensemble techniques to solve the accuracy problem. Our experimental results empirically show that the ensemble approach can reduce bias, and variance, increasing accuracy to achieve the same ReLU performance with parallel and sequential techniques. We demonstrate the effectiveness and robustness of our method using three datasets: MNIST, FMNIST, and CIFAR-10.

KEYWORDS

Homomorphic encryption, activation function, ensemble approach.

Segmentation of Corpus Callosum Using Magnetic Resonance Image and Deep Learning

L. E. Mendoza¹, L. Jaimes², Z. Nieto³, ^{1, 2}Magister, Professor, Research Professor, Faculty of Engineering and Architecture, University of Pamplona, Colombia, ³Doctora, Professor, Research professor, University of Santander, Colombia

ABSTRACT

The automatic detection of specific areas in medical images using mathematical techniques has been growing significantly, due to the applications this allows. This article presents the results in the automatic segmentation of the cerebral corpus callosum in cerebral magnetic resonance imaging using Deep learning. 1450 images were used for training, each image with a resolution of 512*512. A conditioning stage was developed to modify the contrast of the image, remove irrelevant information and perform a pattern extraction process using wavelet transformation. The results show the segmentation of the corpus callosum and the percentage of accuracy was 99.514%. The system was validated with 415 images.

KEYWORDS

Resonance image, cerebral corpus callosum, image segmentation and Deep learning.

Efficient Placement of Mc Nodes and Multicast Routing in an Optical Network

Abdelhakim Dafeur, Samir Ait Belkacem, L Hocine Mouas, Electronics Department, University of Mouloud Mammeri, Tizi-Ouzou, Algeria

ABSTRACT

In this paper, we present two algorithms for allocation of multicast capable nodes “MC nodes” in an all-optical network. We consider the problem of optimally placing a set of MC nodes on a network to minimize the overall links cost and wavelengths number. Simulation results show that our proposed algorithms outperform than the other algorithms significantly. The proposed algorithms are evaluated by the USA and COST-239 topology.

KEYWORDS

Optical Network; Multicast Routing; Wavelength Division Multiplexing; MC nodes Placement; Tree Structure.