Roger Achkar

Online Kernel Adaptive Algorithms With Dictionary Adaptation for MIMO Models

Nonlinear system identification has always been a challenging problem. The use of kernel methods to solve such problems becomes more prevalent. However, the complexity of these methods increases with time which makes them unsuitable for online identification. This drawback can be solved with the introduction of the coherence criterion. Furthermore, dictionary adaptation using a stochastic gradient method proved its efficiency. Mostly, all approaches are used to identify Single Output models which form a particular case of real problems. In this letter we investigate online kernel adaptive algorithms to identify Multiple Inputs Multiple Outputs model as well as the possibility of dictionary adaptation for such models.

Firoxio: Forest fire detection and alerting system

As the human technology moved further, the risk of natural and man induced catastrophes increase exponentially. One of the most dangerous disasters is fires. In addition to its direct danger on humans lives, fire consumes forests where trees that provide humans with oxygen are destroyed. The risk of fire has increased due to the problem of global warming which appeared in the 1980s. Forest fires represent a constant threat to ecological systems, infrastructure and environmental aspects of a community. This gives rise to the urgent need to detect forest fires as fast as possible. This paper highlights the powerful feature of wireless sensor networks as a potential solution to the challenge of early detection of forest fires. The device presented makes use of various sensors attached, solar recharging mechanism, and wireless data transmission, to fulfill the task in question. These collected data are transmitted to a near central unit where they are analyzed and then uploaded to an online website which contacts the Civil Defense unit if necessary. This website is accessible by the specific authorities in order to take early actions in case of any alert. It is worth mentioning that this system is efficient and green; thus, enforcing the need for its creation.

Dictionary adaptation for online prediction of time series data with kernels

Kernel-based algorithms have been a topic of considerable interest in the machine learning community over the last ten years. Their attractiveness resides in their elegant treatment of nonlinear problems. They have been successfully applied to pattern recognition, regression and density estimation. A common characteristic of kernel-based methods is that they deal with kernel expansions whose number of terms equals the number of input data, making them unsuitable for online applications. Recently, several solutions have been proposed to circumvent this computational burden in time series prediction problems. Nevertheless, most of them require excessively elaborate and costly operations. In this paper, we investigate a new model reduction criterion that makes computationally demanding sparsification procedures unnecessary. The increase in the number of variables is controlled by the coherence parameter, a fundamental quantity that characterizes the behavior of dictionaries in sparse approximation problems. We incorporate the coherence criterion into a new kernel-based affine projection algorithm for time series prediction. We also derive the kernel-based normalized LMS algorithm as a particular case. Finally, experiments are conducted to compare our approach to existing methods.

Control of an active magnetic bearing with multi-layer perceptrons using the torque method

The active magnetic bearing (AMB) presents a solution for all the technical problems of the classical bearing since it ensures the total levitation of a body in space eliminating any mechanical contact between the rotor and the stator. The goal of our work is to show the control efficiency of a magnetic sustention, characterized by its nonlinear model, using neural networks (NN). In this paper a study of NN controller for a magnetic bearing under a computed torque control is presented.

Implementation of a Vision System for a Landmine Detecting Robot Using Artificial Neural Network

Landmines, specifically anti-tank mines, cluster bombs, and unexploded ordnance form a serious problem in many countries. Several landmine sweeping techniques are use d for minesweeping. This paper presents the design and the implementation of the vision system of an autonomous robot for landmines localization. The proposed work develops state -of-the-art techniques in digital image processing for pre -processing captured images of the contaminated area. After enhancement, Artificial Neural Network (ANN) is used in order to identify, recognize and classify the landmines’ make and model. The Back-Propagation algorithm is used for training the network. The proposed work prov ed to be able to identify and classify different types of landmines under various conditions (rotated landmine, partially covered landmine) with a success rate of up to 90%.

Simulation of DPCM and ADM Systems

Through years, Digital Communication systems, Pulse Coded Modulation (PCM), Linear Delta Modulation (LDM), Differential Pulse Coded Modulation (DPCM), and Adaptive Delta Modulation (ADM), have proven their unlimited advantages over analog communication systems, in term of error minimization, and distances of transmission enhancement. However two of these systems, the Pulse Coded Modulation and Linear Delta Modulation, still have some weaknesses limiting their advantages, these limitations negatively affect the communication process causing quantization error, slope overload distortion and granular noise. On the other hand, communication engineers have developed two additional digital communication systems which are the Adaptive Delta Modulation (ADM) and the Differential Pulse Coded Modulation (DPCM) in order to solve the aforementioned problems. This paper discusses the implementation and simulation of the aforementioned digital communication systems using Simulink (The Math Works, Inc., Natick, MA, USA) showing the effect of different types of noise when applied to the channel, thus, proving the importance of DPCM and ADM systems in eliminating such effects and ensuring a successful transfer of data.

Remote Generator Control Using Android Application

The traditional technique of monitoring an electric generator was through regular checks on the generators variables: oil, temperature, voltage, and current on a daily basis. Therefore, keeping a normal cycle of performance requires hard work and is often imprecise. The paper presents the solution for the aforementioned issues and more. The idea is to initialize an application that monitors electric generators wireless, using the famous Smart Phone operating system Android. The implemented sensors deliver analog signals that provide real time data about the generators status. These data are converted and programmed though the Arduino microcontroller, which outputs the results in its digital state and then transforms the output into a serial signal, transmitted to the android phone, through a router. Thus a live feedback of the state of the generator is assured. In addition, this project provides a control button that can actually turn this generator on and off. This project is the first step towards the combination of systems and control because it revolutionizes the ideology of monitoring and displaying real time data which can be implemented in various fields depending on different needs. Such fields include electricity, mechanics, and communication. The main limitation faced was the lack of advanced electronics, and technology.

Smart assistive accident free wheelchair system (SAAFWS)

The smart assistive accident free wheel chair system is designed to give a chance for paralyzed patients to move freely within their surroundings. Like the case of any other automated wheelchair, the patient will not exert any effort to move the chair. In the proposed design, the patient will control his chair by simple verbal commands being analyzed by a built-in speech recognition system. Moreover, several additional features are implemented such as heart rate and temperature measurement systems in addition to accident control and obstacle detection systems. All these will be inserted in a webpage in addition to a mobile application to continuously monitor the patients current status. Therefore, the designed wheelchair supports disabled people with secure mobility in addition to continuous vital signs monitoring.

Real-time application of DPCM and ADM systems

Digital communication techniques make the process of modulating a message feasible for transmission. However, a common problem with commonly used techniques, such as Pulse Coded Modulation (PCM) and Linear Delta Modulation (LDM), is that they negatively affect the communication process by causing quantization error, slope overload distortion, and granular noise. This paper discusses the implementation of two modulation systems, Adaptive Delta Modulation (ADM) and Differential Pulse Coded Modulation (DPCM), in order to solve the aforementioned problems. The latter solves the quantization error faced by the PCM and the former solves the slope overload distortion and granular noise faced by LDM. These two systems are implemented using Simulink (The Math Works, Inc., Natick, MA, USA) on a multithreaded processor computer, are tested in real-time, and are subjected to different kinds of noise.

Prediction of Football Matches’ Results Using Neural Networks

In this paper, learning with a teacher artificial neural network to predict the results of football matches is presented. This type of networks requires training via examples, and when the training is complete, the network can be tested to check the results of new examples. In this application, the training examples are the results of previous matches which the network will use to predict the results of new ones.