Browsing by Author "Al-Tam, Faroq"
Now showing 1 - 10 of 17
Results Per Page
Sort Options
- A2PC: augmented advantage pointer-critic model for low latency on mobile IoT with edge computingPublication . Carvalho, Rodrigo; Al-Tam, Faroq; Correia, NoéliaAs a growing trend, edge computing infrastructures are starting to be integrated with Internet of Things (IoT) systems to facilitate time-critical applications. These systems often require the processing of data with limited usefulness in time, so the edge becomes vital in the development of such reactive IoT applications with real-time requirements. Although different architectural designs will always have advantages and disadvantages, mobile gateways appear to be particularly relevant in enabling this integration with the edge, particularly in the context of wide area networks with occasional data generation. In these scenarios, mobility planning is necessary, as aspects of the technology need to be aligned with the temporal needs of an application. The nature of this planning problem makes cutting-edge deep reinforcement learning (DRL) techniques useful in solving pertinent issues, such as having to deal with multiple dimensions in the action space while aiming for optimum levels of system performance. This article presents a novel scalable DRL model that incorporates a pointer-network (Ptr-Net) and an actor-critic algorithm to handle complex action spaces. The model synchronously determines the gateway location and visit time. Ultimately, the gateways are able to attain high-quality trajectory planning with reduced latency.
- Adaptive spectrum allocation for 5G wireless communication scenariosPublication . Correia, Noélia; Al-Tam, Faroq; Rodriguez, J.5G resources should be properly planned for users to have a good quality of service. This planning includes defining the most suitable numerology indexes and best spectrum allocation considering the requirements of current traffic, which may change over time. Furthermore, when accounting for changes in traffic pattern, any necessary reconfigurations should be minimized. Here in this article, an optimization model is developed for the planning of spectrum allocation to the best mix numerology. The model considers adapting the operating numerology mix according to the current presence (or not) of traffic requirements. The model also works under any wireless communication scenario in 5G, and under any traffic pattern.
- Analysis of machine learning techniques applied to sensory detection of vehicles in intelligent crosswalksPublication . Lozano Domínguez, José Manuel; Al-Tam, Faroq; Mateo Sanguino, Tomás de J.; Correia, NoéliaImproving road safety through artificial intelligence-based systems is now crucial turning smart cities into a reality. Under this highly relevant and extensive heading, an approach is proposed to improve vehicle detection in smart crosswalks using machine learning models. Contrarily to classic fuzzy classifiers, machine learning models do not require the readjustment of labels that depend on the location of the system and the road conditions. Several machine learning models were trained and tested using real traffic data taken from urban scenarios in both Portugal and Spain. These include random forest, time-series forecasting, multi-layer perceptron, support vector machine, and logistic regression models. A deep reinforcement learning agent, based on a state-of-the-art double-deep recurrent Q-network, is also designed and compared with the machine learning models just mentioned. Results show that the machine learning models can efficiently replace the classic fuzzy classifier.
- Attention-based model and deep reinforcement learning for distribution of event processing tasksPublication . Mazayev, Andriy; Al-Tam, Faroq; Correia, NoéliaEvent processing is the cornerstone of the dynamic and responsive Internet of Things (IoT). Recent approaches in this area are based on representational state transfer (REST) principles, which allow event processing tasks to be placed at any device that follows the same principles. However, the tasks should be properly distributed among edge devices to ensure fair resources utilization and guarantee seamless execution. This article investigates the use of deep learning to fairly distribute the tasks. An attention-based neural network model is proposed to generate efficient load balancing solutions under different scenarios. The proposed model is based on the Transformer and Pointer Network architectures, and is trained by an advantage actorcritic reinforcement learning algorithm. The model is designed to scale to the number of event processing tasks and the number of edge devices, with no need for hyperparameters re-tuning or even retraining. Extensive experimental results show that the proposed model outperforms conventional heuristics in many key performance indicators. The generic design and the obtained results show that the proposed model can potentially be applied to several other load balancing problem variations, which makes the proposal an attractive option to be used in real-world scenarios due to its scalability and efficiency.
- Deep PC-MAC: a deep reinforcement learning pointer-critic media access protocolPublication . Al-Tam, Faroq; Mazayev, Andriy; Correia, Noélia; Rodriguez, J.Developing artificial intelligence (AI) solutions for communication problems is one of the hottest topics nowadays. This article presents Deep PC-MAC, a novel deep reinforcement learning (DRL) solution to solve the fair coexistence problem (FCP) between heterogeneous nodes in the unlicensed bands. It is based on a hybrid architecture between pointer networks (Ptr-nets) and advantage actor-critic (A2C), i.e., pointer-critic architecture. The proposed model allows base stations to fairly share unlicensed bands with incumbent nodes. It jointly protects the incumbent nodes from spectrum starvation and improves key-performance indicators (KPIs). Deep PC-MAC is trained from scratch with zero-knowledge about FCP and experimental results demonstrate its efficiency when compared to a baseline method.
- DNAGear: a free software for spa type identification in Staphylococcus aureusPublication . Al-Tam, Faroq; Brunel, Anne-Sophie; Bouzinbi, Nicolas; Corne, Philippe; Bañuls, Anne-Laure; Shahbazkia, Hamid RezaStaphylococcus aureus is both human commensal and an important human pathogen, responsible for community-acquired and nosocomial infections ranging from superficial wound infections to invasive infections, such as osteomyelitis, bacteremia and endocarditis, pneumonia or toxin shock syndrome with a mortality rate up to 40%. S. aureus reveals a high genetic polymorphism and detecting the genotypes is extremely useful to manage and prevent possible outbreaks and to understand the route of infection. One of current and expanded typing method is based on the X region of the spa gene composed of a succession of repeats of 21 to 27 bp. More than 10000 types are known. Extracting the repeats is impossible by hand and needs a dedicated software. Unfortunately the only software on the market is a commercial program from Ridom. Findings This article presents DNAGear, a free and open source software with a user friendly interface written all in Java on top of NetBeans Platform to perform spa typing, detecting new repeats and new spa types and synchronizing automatically the files with the open access database. The installation is easy and the application is platform independent. In fact, the SPA identification is a formal regular expression matching problem and the results are 100% exact. As the program is using Java embedded modules written over string manipulation of well established algorithms, the exactitude of the solution is perfectly established. Conclusions DNAGear is able to identify the types of the S. aureus sequences and detect both new types and repeats. Comparing to manual processing, which is time consuming and error prone, this application saves a lot of time and effort and gives very reliable results. Additionally, the users do not need to prepare the forward-reverse sequences manually, or even by using additional tools. They can simply create them in DNAGear and perform the typing task. In short, researchers who do not have commercial software will benefit a lot from this application.
- Flow setup aware controller placement in distributed software-defined networkingPublication . Correia, Noélia; Al-Tam, FaroqIn distributed software-defined networking, switches can be assigned to a single master controller and one or more slave controllers, for resilience. However, only master controllers are allowed to install flow rules. Also, controllers install flow rules just at switches under their domain. This means that a particular flow may trigger the flow setup procedure multiple times, if the traversed switches are under different controller domains. This results into extra network load and instability. Here, in this article, the controller placement is planned having into account flow setup efficiency and possible future migrations from master to slave controllers. Results show that such flow setup and migration awareness results into long-term quality solutions for controller placement.
- Fractional switch migration in multi-controller software-defined networkingPublication . Al-Tam, Faroq; Correia, NoéliaMapping switches to controllers in multi-controller software-defined networking (SDN) is still a hot research topic. Many factors have to be considered when establishing this mapping. Among them are the load balancing and mapping stability. Load balancing is important to improve resources utilization, and mapping stability reduces the control plane overhead created when exchanging information triggered by new mappings. This article presents a model for dynamic switch-controller mapping to achieve load balancing and minimize the number of new switch-controller assignments. To that end, for load balancing, flows from a switch are allowed to be handled by multiple controllers, and to increase assignment stability, the assignments at time t - 1 are taken into consideration when calculating the assignments at time t. The model is formulated as a convex quadratic programming problem, and the properties and feasibility of this model are mathematically analyzed. In addition, a heuristic algorithm is developed to deal with large-scale networks. The experimental results show the effectiveness of the proposed approach when compared to recent academic work, where the proposed model leads to a slight improvement in the load balancing and increases the stability of the switch-controller assignment by approximate to 91%. (C) 2019 Elsevier B.V. All rights reserved.
- Hessian-polar context: a descriptor for microfilaria recognitionPublication . Al-Tam, Faroq; dos Anjos, Antonio; Shahbazkia, Hamid R.This paper presents a new effective descriptor for microfilaria. Since microfilaria is a thin elastic object, the proposed descriptor handles it locally. At each medial point of the microfilaria, the local structure of the microfilaria votes for a given shape. Accumulating these votes in the polar domain yields a rich descriptor. Experimental results show the effectiveness of the proposed approach when compared to a set of different well-established methods.
- Illumination correction and analysis of two-dimensional microscopy images of Loa loa microfilariaePublication . Al-Tam, Faroq; Shahbazkia, Hamid Reza; Anjos, António dosThis thesis addresses the problem of detecting a common parasitic micro laria that causes loaisis, a major disease problem in Central and Western Africa. The dose of medicine to be administered to the patient is proportional to the estimated number of micro lariae in the patient's body. Therefore, proper estimation of the number of micro lariae is the key for conducting the right procedure. The clinical examination is necessary to estimate the micro lariae density in a blood sample drawn from the patient. Thereafter, visual inspection of the sample is performed. The main challenge in this work is, however, the development of an automatic detection system of micro lariae in 2-D images. Such problem is new in the image processing literature, and the development of such system is very important for performing better diagnosis and treatment of this disease and other similar diseases. A comprehensive review of, both generic and thin, object detectors in 2-D images is presented. A very robust method for microscopy image illumination correction is proposed, and a new powerful descriptor, the Hessian-Polar Context (HPC), for micro lariae is also introduced. These are then combined in a micro lariae detection system, where a simple, yet e cient, hypotheses generator is also presented. Additionally, several methods and applications for di erent image modalities are proposed. These involve a method and an application for the analysis of rice panicle in 2-D images. Additionally, an e cient method for artifact suppression in X-ray image is also proposed. The proposed methods are compared to a set of state-of-the-art methods. Experimental results show that the developed methods are great contributions to the microscopy and X-ray imaging elds.