Papers by Prof. Hussein CHIBLE
East Asian Journal of Multidisciplinary Research
To maintain the benefits and improve the course procedure for the next year, this work is devoted... more To maintain the benefits and improve the course procedure for the next year, this work is devoted to evaluating my course of computer at the faculty of tourism at the Lebanese University during the academic year 2021–2022. Additionally, we want to mention that this method can be used as a model for any similar course evaluation. According to the results, some variables of grades are normally distributed while others are not normally distributed. The final grade average and partial grade average for each student are different from one another. Final grades were statistically equal for both male and female students. The final grade average for this year and last year is the same. There is a correlation between participation and presence grades and the other grades, except the mid-exam grades. Also, the final grades and all other grades, except the mid-exam grades, are correlated. For the following year, we need to make sure that students attend more lectures to raise their grades. Als...
Abstract- The hyperbolic tangent function and its derivative are key essential element in analog ... more Abstract- The hyperbolic tangent function and its derivative are key essential element in analog signal processing and especially in analog VLSI implementation of neuron of artificial neural networks. The main conditions of these types of circuits are the small silicon area, and the low power consumption. The objective of this paper is to study and design CMOS VLSI hyperbolic tangent function and its derivative circuit for neural network implementation. A circuit is designed and the results are presented
Lecture Notes in Electrical Engineering, 2021
With the rise of Internet of Things (IoT) and Edge Computing, which are technologies that rely on... more With the rise of Internet of Things (IoT) and Edge Computing, which are technologies that rely on smart and low power computing nodes with adequate processing power and storage capabilities, it is expected that Artificial Intelligence and machine learning will play a role in the continuous spreading of their application fields. One of the most adopted hardware platforms for IoT and Machine Learning is the low-cost, multipurpose Raspberry Pi, which is small enough and still capable of effectively handling machine learning tasks. Moreover, it is ideal for development and educational purposes. On the other hand, among the plethora of Machine Learning (ML) paradigms reported in the literature, we identified Rulex [1] as a good candidate as an ML engine, suitable for advanced edge computing applications. In this paper, we report the deployment of the machine learning package Rulex to operate on the Raspberry Pi in multiple arrangements. The target is to perform training and testing of Machine Learning algorithms through running Rulex on the Raspberry PI as an Edge Computing Device. Specifically, we describe the process of porting Rulex external and internal libraries on Windows 32 Bits, Ubuntu 64 Bits, and Raspbian 32 Bits. Moreover, we present the standalone and Client/Server Configuration of Rulex on the Raspberry Pi along with the Remote Development configuration used to compile and debug the Rulex source code remotely. We have applied Forecasts using training and testing data sets on the Raspberry Pi as an IoT Device, which generate promising and accurate results
International Journal of Modelling and Simulation, 2015
Abstract In this paper, a new implementation of CMOS four-quadrant analogue synapse multiplier ci... more Abstract In this paper, a new implementation of CMOS four-quadrant analogue synapse multiplier circuit for analogue signal processing will be proposed. Especially, it can be used for multi-layer perceptron neural networks. The proposed multiplier is composed of only four transistors and it will multiply two input currents and produces an output current. The global multiplier circuit consists of 10 transistors, but only four of them will be implemented inside the synapse, while the others will be implemented inside the input and the neuron. The main characteristics of the proposed circuit are the small silicon area and the low power consumption. A comparison among some other multipliers will be presented.
International Journal of Computer Applications Technology and Research, 2017
Gas sensing in biomedical applications shows a conversion of the concentration of the exhaled gas... more Gas sensing in biomedical applications shows a conversion of the concentration of the exhaled gas to a variation in resistance, so an electronic integrated interface circuit is required to analyse the exhaled gases, which are indications for many diseases. In this paper, a resistance to current conversion circuit based on differential biasing for Electronic nose (E-nose) breath analyser is presented. Over more than 5-decades (500Ω to 100MΩ) input resistance range, a precision, less than 1%, required by novel gas sensing system in portable applications, is preserved. Therefore, the proposed circuit obtains high accuracy under simulation. The outputs of the proposed Resistance to Current (R-to-I) conversion circuit achieve a percentage error below 0.25% under environment corners. The reliability of the proposed circuit is also investigated under the effect of process variations. In order to assess the correctness of the proposed architecture, the circuit was compared to similar solutions presented in literature where the proposed architecture attains a worst-case percentage error of 0.05%.
Analog VLSI Neural Networks are massively parallel analog systems, which demonstrated capabilitie... more Analog VLSI Neural Networks are massively parallel analog systems, which demonstrated capabilities in solving a wide range of real world problems: as a consequence analog neural computing techniques have became widespread [1]. In this context, analog CMOS electronic circuits are key computational elements and it is very important to efficiently implement them. The Neural Network proposed is the Multi Layer Perceptron and the learning algorithm is the Back Propagation. The neural network first is implemented by using Software "C Language". This network is used - for the first experiment - to learn patterns such as numbers (0,1,2…9). The patterns are real; 2000 handwritten patterns (200 Zeros; 200 Ones;…..200 Nines) are divided into two sets: the training set (1000) and the test set (1000). The Neural Network uses the training set to learn the numbers and to update itself to understand them. The test set is used to test the neural network generalization. The Neural Network m...
Abstract: In this paper I present a four quadrant analog multiplier. The multiplier can be used f... more Abstract: In this paper I present a four quadrant analog multiplier. The multiplier can be used for general purpose but it is suitable for Analog VLSI implementation of artificial neural networks because of its small silicon area and its low power consumption. The main feature of the multiplier is the high value of the weight voltage range, it can take values between the ground voltage and the supply volt-age [0:Vdd]. The innovative idea in the multiplier proposed is to obtain a linear function by substituting quadratic function inside the square root function.
In this paper, I present a method for students' exams by using computers. A computer program ... more In this paper, I present a method for students' exams by using computers. A computer program developed to correct the exams automatically and the system used to support the exam method are explained. This method is proved and tested in our faculty. Practical examples of exams are verified.
2015 Third International Conference on Technological Advances in Electrical, Electronics and Computer Engineering (TAEECE), 2015
Artificial neural networks aim to simulate the human nervous system using the interprocessing cal... more Artificial neural networks aim to simulate the human nervous system using the interprocessing calculation methodology, but unfortunately cannot the preserve stimulus pattern. In this paper, we depend on the biological fact “information is coded within firing rate” and hence we propose an architecture for the preceptor in which the neurons' APs (Action Potentials) are transmitted in structures that represent the stimuli patterns, and the response of connected neuron through their synapses is highly proportional to the nature of these structures. The new preceptor that uses vector in space as input and the magic dyadic matrix shows a significant enhancement in many factors.
Computing
Machine learning techniques aim to mimic the human ability to automatically learn how to perform ... more Machine learning techniques aim to mimic the human ability to automatically learn how to perform tasks through training examples. They have proven capable of tasks such as prediction, learning and adaptation based on experience and can be used in virtually any scientific application, ranging from biomedical, robotic, to business decision applications, and others. However, the lack of domain knowledge for a particular application can make feature extraction ineffective or even unattainable. Furthermore, even in the presence of pre-processed datasets, the iterative process of optimizing Machine Learning parameters, which do not translate from one domain to another, maybe difficult for inexperienced practitioners. To address these issues, we present in this paper a Vectorized Automated ML Pre-processIng and post-pRocEssing framework, approximately named (VAMPIRE), which implements feature extraction algorithms capable of converting large time-series recordings into datasets. Also, it i...
2017 29th International Conference on Microelectronics (ICM)
This paper presents a motion sensing add-on and a graphical user interface for the balance board.... more This paper presents a motion sensing add-on and a graphical user interface for the balance board. Our system adds kinetic sensing to the board in order to record its rotation in the 3D space. A sports game, a 3D video game for ski training, is set to be controlled by the rotation of the balance board. The system thus adds to the balance board IoT capabilities.
2017 New Generation of CAS (NGCAS), 2017
2019 26th IEEE International Conference on Electronics, Circuits and Systems (ICECS), 2019
Kalman filter is designed to predict the correct values for noisy measurements, using a probabili... more Kalman filter is designed to predict the correct values for noisy measurements, using a probabilistic model. However, the chosen system model has a huge impact on the outcome of this algorithm, which makes it very important for the model to be as representative of the system as possible. A new approach to handle the noisy measurements in low power Maximum Power Point Tracking (MPPT) algorithms is presented in this paper. The proposed filtering proved beneficial in terms of algorithm stability.
2017 13th Conference on Ph.D. Research in Microelectronics and Electronics (PRIME), 2017
In this paper, we present a new version of previously fabricated event driven tactile sensor [1] ... more In this paper, we present a new version of previously fabricated event driven tactile sensor [1] with modifications across its circuits and methods involved using the AMS CMOS 0.18μm technology. Electrical characterization experimental results are shown and compared to the ones from the previous proving the advantages of the new sensor concerning area saving, sensing small input signals and power consumption.
The operation of modern electronic devices in different fields as communications, signal processi... more The operation of modern electronic devices in different fields as communications, signal processing, and sensor interface is critically affected with robust, high performance and scalable Analog-to-Digital Converter (ADCs), that can be considered as one of the main blocks in many systems, since they are mandatory to make the link between the analog outside world and the evermore-ubiquitous digital computer world. The design of these ADCs come distinct tradeoffs between speed, power, resolution, and die area embodied within many data conversion architectural variations. The flash ADC structure are often the base structure for high-speed operation and simple architecture analog-to-digital converters (ADCs). As the input signal is applied to (2
2020 32nd International Conference on Microelectronics (ICM), 2020
Detection and classification of moving targets is an essential feature in many applications like ... more Detection and classification of moving targets is an essential feature in many applications like road surveillance systems, autonomous cars, and smart gate systems. Multi-chirp sequence Frequency Modulated Continuous Wave (FMCW) radars with a 2D FFT processing can be used to produce a Range-Doppler images (R-D maps) containing the signature of the target. However, in low-cost FMCW radars, these images suffer from many problems like low-resolution and ambiguity. Such problems can make the image look unrealistic as well as hard to process and classify. In this paper, we propose a human-vehicle classification method based on Transfer Learning. The classification is done by processing the R-D maps generated by a low-cost short range 24 GHz FMCW radar with a convolutional Neural Network (CNN). The adopted CNN succeeded to reach a 96.5% accuracy in discriminating humans from vehicles.
Uploads
Papers by Prof. Hussein CHIBLE