Search

International audience; Recent development in smart devices has lead us to an explosion in data generation and heterogeneity, which requires new network solutions for better analysing and understanding traffic. These solutions should be intelligent and scalable in order to handle the huge amount of data automatically. With the progress of high-performance computing (HPC), it becomes feasible easily to deploy machine learning (ML) to solve complex problems and its efficiency has been validated in several domains (e.g., healthcare or computer vision). At the same time, network slicing (NS) has drawn significant attention from both industry and academia as it is essential to address the diversity of service requirements. Therefore, the adoption of ML within NS management is an interesting issue. In this paper, we have focused on analyzing network data with the objective of defining network slices according to traffic flow behaviors. For dimensionality reduction, the feature selection has been applied to select the most relevant features (15 out of 87 features) from a real dataset of more than 3 million instances. Then, a K-Means clustering is applied to better understand and distinguish behaviors of traffic. The results demonstrated a good correlation among instances in the same cluster generated by the unsupervised learning. This solution can be further integrated in a real environment using network function virtualization.

International audience; Recent medical applications are largely dominated by the application of Machine Learning (ML) models to assist expert decisions, leading to disruptive innovations in radiology, pathology, genomics, and hence modern healthcare systems in general. Despite the profitable usage of AI-based algorithms, these data-driven methods are facing issues such as the scarcity and privacy of user data, as well as the difficulty of institutions exchanging medical information. With insufficient data, ML is prevented from reaching its full potential, which is only possible if the database consists of the full spectrum of possible anatomies, pathologies, and input data types. To solve these issues, Federated Learning (FL) appeared as a valuable approach in the medical field, allowing patient data to stay where it is generated. Since an FL setting allows many clients to collaboratively train a model while keeping training data decentralized, it can protect privacy-sensitive medical data. However, FL is still unable to deliver all its promises and meets the more stringent requirements (e.g., latency, security) of a healthcare system based on multiple Internet of Medical Things (IoMT). For example, although no data are shared among the participants by definition in FL systems, some security risks are still present and can be considered as vulnerabilities from multiple aspects. This paper sheds light upon the emerging deployment of FL, provides a broad overview of current approaches and existing challenges, and outlines several directions of future work that are relevant to solving existing problems in federated healthcare, with a particular focus on security and privacy issues.

COVID-19 pandemic has become endemic and has plunged the global community into a perilous situation pervaded with an economic recession, loss of jobs, and the death of thousands of people. It spreads exponentially around the world, affects 213 countries and territories as well as two international conveyances. Yet, the pandemic has neither clinically proven drugs nor vaccines. Therefore, it is now evident that non-medical approaches such as deep learning, data mining, expert system, software agents, and other artificial intelligence techniques are urgently needed to combat the pandemic, provide alternative solutions to alleviate the huge burden on the limited health care systems available around the world and curtail the future outbreak of the COVID-19 pandemic. Specifically, deep learning (DL) techniques evolved from machine learning (ML) concepts over a period of time and have been amply embraced in many real-life applications because of its unique nature and features for solving problems. Moreover, it is a powerful method of data exploration, and more importantly, has outperformed human efforts in several areas such as computer vision and health-related applications. Therefore, DL can be employed for combating and mitigating the proliferation of COVID-19 virus among humans. This chapter introduces the concept of deep learning and its potentials for combating the current spread COVID-19 pandemic and mitigating future outbreaks, discussed ongoing efforts of deep learning as one of the non-clinical approaches to alleviate the spread and curtail the further outbreak COVID-19 pandemic as well as the challenges of deep learning in combating COVID-19 pandemic and future directions.

Accepted in IEEE ICC 2021 Workshop on Workshop on Data Driven Intelligence for Networks and Systems (DDINS); The rapid development of smart devices triggers a surge in new traffic and applications. Thus, network traffic classification has become a challenge in modern communications and may be applied to a various range of applications ranging from QoS provisioning to security-related applications. Developing Machine Learning (ML) methods, which can successfully distinguish network applications from each other, is one of the most important tasks. Since ML algorithms are as good as the quality of data, feature selection has become a crucial step in the ML process. Therefore, selecting effective and relevant features for traffic analysis is also another essential issue. In this paper, we are interested in identifying the most relevant features to characterize network traffic. Empirical results indicate that significant input feature selection is important to classify network traffic. Then, a comparative analysis of various Decision Treebased models (both traditional and recent algorithms) has been conducted with feature selection methods in terms of accuracy, training, and classification time.

Ambient Intelligence (AmI) is a new paradigm that redefines interaction between humans, sensors and flow of data and information. In AmI, environment is more sensitive and more responsive to the user who acts spontaneously in the foreground, while sensors, machines and intelligent methods act in background. Behind the AmI interfaces, a huge volume of data is collected and analysed to make decision in real time. In the field of health care, AmI solutions prevent the patients from emergency situations by using data mining techniques. This paper assess the performance of deep learning against traditional dimensionality reduction and classification algorithms for healthcare monitoring in a hospital environment. An ensemble method is proposed using three classifiers on a reduced version of the dataset, where the dimensionality reduction technique is based on deep learning. The evaluation based on different performance metrics like accuracy, precision, recall and f-measure illustrated reliable performance of the proposed ensemble method.

O Aouedi, K Piamrat, D Bagadthey, "Handling partially labeled network data: a semi-supervised approach using stacked sparse autoencoder", Accepted in Computer Networks, Elsevier; International audience; Network traffic analytics has become a crucial task in order to better understand and manage network resources, especially in the network softwarization era where the implementation of this concept can be done easily with network function virtualization. Currently, many approaches have been proposed to improve the performance of traffic classification. However, as new types of traffic emerge every day (and they are generally not labeled), this opens a new challenge to be handled. Moreover, the question of how to accurately classify traffic using a limited amount of labeled data or partially labeled data hence becomes another important concern. In fact, labeling data is often difficult and time-consuming. In order to tackle the previously described issues, we reformulate traffic classification into a semi-supervised learning where both supervised learning (using labeled data) and unsupervised learning (no label data) are combined. To do so, this paper presents a stacked sparse autoencoder (SSAE) based semi-supervised deep-learning model for traffic classification. The main motivations of this approach are: (i) unlabeled data is often abundant and easily available; (ii) classification performance of the whole model can be greatly improved when a large amount of unlabeled traffic is included in the training process; (iii) there is a limit to how much human effort can be thrown at the labeling problem. To investigate the performance of our approach, an empirical study has been conducted on a real dataset and results indicate that using a large amount of unlabeled data in the SSAE pre-trained phase can improve significantly the classification performance of the whole model. The proposed approach is compared against other representative machine-learning and deep-learning models, which are Support Vector Machine (SVM), Decision Tree (DT), Random Forest (RF), Multi-Layer Perceptron (MLP), eXtreme Gradient Boosting (XGBoost), and Autoencoder. Furthermore, we have also conducted experiments on a well-known dataset including encrypted traffic (containing only time-related features) to evaluate the generalization performance of the proposed model.

Accepted for publication in IEEE ICC 2022; International audience; With the increasing development of 5G/Beyond 5G and network softwarization techniques, we have more flexibility and agility in the network. This can be exploited by Machine Learning (ML) to integrate intelligence in the network and improve network as well as service management in edge-cloud environment. Intrusion detection systems (IDS) is one of the challenging issues for managing network. However, traditional approaches in this domain require all data (and their associated labels) to be centralized at the same location. In this context, such approaches lead to: (i) a large bandwidth overhead, as raw data needs to be transmitted to the server, (ii) low incentives for devices to send their private data, and (iii) large computing and storage resources needed on the server side to label and treat all this data. In this paper, to cope with the above limitations, we propose a semi-supervised federated learning model for IDS. Moreover, we use network softwarisation for automation and deployment. Our model combines Federated Learning and Semi-Supervised Learning where the clients train unsupervised models (using unlabeled data) to learn the representative and low-dimensional features and the server conducts a supervised model (using labeled data). We evaluate this approach on the well-known UNSW-NB15 dataset and the experimental results demonstrate that our approach can achieve accuracy and detection rates up to 84.32% and 83.10%, respectively while keeping the data private with limited overhead.