Federated Learning Approach for Secured Medical Recommendation in Internet of Medical Things Using Homomorphic Encryption.

The concept of Federated Learning (FL) is a distributed-based machine learning (ML) approach that trains its model using edge devices. Its focus is on maintaining privacy by transmitting gradient updates along with users' learning parameters to the global server in the process of training as well as preserving the integrity of data on the user-end of internet of medical things (IoMT) devices. Instead of a direct use of user data, the training which is performed on the global server is done on the parameters while the model modification is performed locally on IoMT devices. But the major drawback of this federated learning approach is its inability to preserve user privacy complete thereby resulting in gradients leakage. Thus, this study first presents a summary of the process of learning and further proposes a new approach for federated medical recommender system which employs the use of homomorphic cryptography to ensure a more privacy-preservation of user gradients during recommendations. The experimental results indicate an insignificant decrease with respect to the metrics of accuracy, however, a greater percentage of user-privacy is achieved. Further analysis also shows that performing computations on encrypted gradients at the global server scarcely has any impact on the output of the recommendation while guaranteeing a supplementary secure channel for transmitting user-based gradients back and forth the global server. The result of this analysis indicates that the performance of federated stochastic modification minimized gradient (FSMMG) algorithm is greatly increased at every given increase in the number of users and a good convergence is achieved as well. Also, experiments indicate that when compared against other existing techniques, the proposed FSMMG outperforms at 98.3% encryption accuracy.

Blockchain-Secured Recommender System for Special Need Patients Using Deep Learning.

Recommender systems offer several advantages to hospital data management units and patients with special needs. These systems are more dependent on the extreme subtle hospital-patient data. Thus, disregarding the confidentiality of patients with special needs is not an option. In recent times, several proposed techniques failed to cryptographically guarantee the data privacy of the patients with special needs in the diet recommender systems (RSs) deployment. In order to tackle this pitfall, this paper incorporates a blockchain privacy system (BPS) into deep learning for a diet recommendation system for patients with special needs. Our proposed technique allows patients to get notifications about recommended treatments and medications based on their personalized data without revealing their confidential information. Additionally, the paper implemented machine and deep learning algorithms such as RNN, Logistic Regression, MLP, etc., on an Internet of Medical Things (IoMT) dataset acquired via the internet and hospitals that comprises the data of 50 patients with 13 features of various diseases and 1,000 products. The product section has a set of eight features. The IoMT data features were analyzed with BPS and further encoded prior to the application of deep and machine learning-based frameworks. The performance of the different machine and deep learning methods were carried out and the results verify that the long short-term memory (LSTM) technique is more effective than other schemes regarding prediction accuracy, precision, F1-measures, and recall in a secured blockchain privacy system. Results showed that 97.74% accuracy utilizing the LSTM deep learning model was attained. The precision of 98%, recall, and F1-measure of 99% each for the allowed class was also attained. For the disallowed class, the scores were 89, 73, and 80% for precision, recall, and F1-measure, respectively. The performance of our proposed BPS is subdivided into two categories: the secured communication channel of the recommendation system and an enhanced deep learning approach using health base medical dataset that spontaneously identifies what food a patient with special needs should have based on their disease and certain features including gender, weight, age, etc. The proposed system is outstanding as none of the earlier revised works of literature described a recommender system of this kind.

Realizing Efficient Security and Privacy in IoT Networks.

In recent times, security and privacy at the physical (PHY) layer has been a major issue of several communication technologies which comprise the internet of things (IoT) and mostly, the emerging fifth-generation (5G) cellular network. The most real-world PHY security challenge stems from the fact that the passive eavesdropper's information is unavailable to the genuine source and destination (transmitter/receiver) nodes in the network. Without this information, it is difficult to optimize the broadcasting parameters. Therefore, in this research, we propose an efficient sequential convex estimation optimization (SCEO) algorithm to mitigate this challenge and improve the security of physical layer (PHY) in a three-node wireless communication network. The results of our experiments indicate that by using the SCEO algorithm, an optimal performance and enhanced convergence is achieved in the transmission. However, considering possible security challenges envisaged when a multiple eavesdropper is active in a network, we expanded our research to develop a swift privacy rate optimization algorithm for a multiple-input, multiple-output, multiple-eavesdropper (MIMOME) scenario as it is applicable to security in IoT and 5G technologies. The result of the investigation show that the algorithm executes significantly with minimal complexity when compared with nonoptimal parameters. We further employed the use of rate constraint together with self-interference of the full-duplex transmission at the receiving node, which makes the performance of our technique outstanding when compared with previous studies.

The Use of Ensemble Models for Multiple Class and Binary Class Classification for Improving Intrusion Detection Systems.

The pursuit to spot abnormal behaviors in and out of a network system is what led to a system known as intrusion detection systems for soft computing besides many researchers have applied machine learning around this area. Obviously, a single classifier alone in the classifications seems impossible to control network intruders. This limitation is what led us to perform dimensionality reduction by means of correlation-based feature selection approach (CFS approach) in addition to a refined ensemble model. The paper aims to improve the Intrusion Detection System (IDS) by proposing a CFS + Ensemble Classifiers (Bagging and Adaboost) which has high accuracy, high packet detection rate, and low false alarm rate. Machine Learning Ensemble Models with base classifiers (J48, Random Forest, and Reptree) were built. Binary classification, as well as Multiclass classification for KDD99 and NSLKDD datasets, was done while all the attacks were named as an anomaly and normal traffic. Class labels consisted of five major attacks, namely Denial of Service (DoS), Probe, User-to-Root (U2R), Root to Local attacks (R2L), and Normal class attacks. Results from the experiment showed that our proposed model produces 0 false alarm rate (FAR) and 99.90% detection rate (DR) for the KDD99 dataset, and 0.5% FAR and 98.60% DR for NSLKDD dataset when working with 6 and 13 selected features.