Mohsen Guizani

Bio: Mohsen Guizani is an academic researcher from Qatar University. The author has contributed to research in topics: Computer science & Cloud computing. The author has an hindex of 79, co-authored 1110 publications receiving 31282 citations. Previous affiliations of Mohsen Guizani include Jaypee Institute of Information Technology & University College for Women.

Deep Multi-Layer Perceptron Classifier for Behavior Analysis to Estimate Parkinson’s Disease Severity Using Smartphones

Abstract: Although the preclinical detection of Parkinson’s disease (PD) has been explored, a practical, inexpensive, and overall screening diagnosis has yet to be made available. However, due to the large variability and complexity in progress of PD and the difficulties in gathering a single time-point measurement of a single sign, the goal of precision treatment and assessment severity would be impossible to achieve. Hence, the repeated monitoring and tracking of individuals during their daily living activities at different times would also be of great importance for treating this chronic disease. We propose a deep multi-layer perceptron (DMLP) classifier for behavior analysis to estimate the progression of PD using smartphones. This paper aims to identify severity in PD patients’ actions by analyzing their speech and movement patterns, as measured with a smartphone accelerometer in their pocket at different times of the day. Popular machine learning classification algorithms, such as logistic regression, random forests, k-nearest neighbors, M5P, and DMLP, are applied on one dataset from the University of California Irvine and another dataset collected by the authors to classify each patient as being Parkinson positive or negative. We further measure the success of each method for their ability to correctly classify the patients into one of these categories. Of the experimental models, it is demonstrated that DMLP performs the best in both datasets.

A Review on the Role of Machine Learning in Enabling IoT Based Healthcare Applications

Abstract: The Internet of Things (IoT) is playing a vital role in the rapid automation of the healthcare sector. The branch of IoT dedicated towards medical science is at times termed as Healthcare Internet of Things (H-IoT). The key elements of all H-IoT applications are data gathering and processing. Due to the large amount of data involved in healthcare, and the enormous value that accurate predictions hold, the integration of machine learning (ML) algorithms into H-IoT is imperative. This paper aims to serve both as a compilation as well as a review of the various state of the art applications of ML algorithms currently being integrated with H-IoT. Some of the most widely used ML algorithms have been briefly introduced and their use in various H-IoT applications has been analyzed in terms of their advantages, scope, and possible improvements. Applications have been divided into the domains of diagnosis, prognosis and spread control, assistive systems, monitoring, and logistics. In healthcare, practical use of a model requires it to be highly accurate and to have ample measures against security attacks. The applications of ML algorithms in H-IoT discussed in this paper have shown experimental evidence of accuracy and practical usability. The constraints and drawbacks of each of these applications have also been described.

Reinforcement learning for resource provisioning in the vehicular cloud

Abstract: This article presents a concise view of vehicular clouds that incorporates various vehicular cloud models that have been proposed to date. Essentially, they all extend the traditional cloud and its utility computing functionalities across the entities in the vehicular ad hoc network. These entities include fixed roadside units, onboard units embedded in the vehicle, and personal smart devices of drivers and passengers. Cumulatively, these entities yield abundant processing, storage, sensing, and communication resources. However, vehicular clouds require novel resource provisioning techniques that can address the intrinsic challenges of dynamic demands for the resources and stringent QoS requirements. In this article, we show the benefits of reinforcement-learning-based techniques for resource provisioning in the vehicular cloud. The learning techniques can perceive long-term benefits and are ideal for minimizing the overhead of resource provisioning for vehicular clouds.

Capacity of Hybrid Wireless Networks with Directional Antenna and Delay Constraint

Abstract: We study the throughput capacity of hybrid wireless networks with a directional antenna. The hybrid wireless network consists of n randomly distributed nodes equipped with a directional antenna, and m regularly placed base stations connected by optical links. We investigate the ad hoc mode throughput capacity when each node is equipped with a directional antenna under an L-maximum-hop resource allocation. That is, a source node transmits to its destination only with the help of normal nodes within L hops. Otherwise, the transmission will be carried out in the infrastructure mode, i.e., with the help of base stations. We find that the throughput capacity of a hybrid wireless network greatly depends on the maximum hop L, the number of base stations m, and the beamwidth of directional antenna θ. Assuming the total bandwidth W bits/sec of the network is split into three parts, i.e., W1 for ad hoc mode, W2 for uplink in the infrastructure mode, and W3 for downlink in the infrastructure mode. We show that the throughput capacity of the hybrid directional wireless network is Θ(nW1/θ2 L log n) + Θ(mW2), if L = Ω(n1/3/θ4/3 log2/3 n); and Θ((θ2L2 log2/3 n); and + Θ(mW2), if L = o(n1/3/θ4/3 log2/3 n), respectively. Finally, we analyze the impact of L, m and θ on the throughput capacity of the hybrid networks.

Deep CNN-Based Real-Time Traffic Light Detector for Self-Driving Vehicles

Abstract: Due to the unavailability of Vehicle-to-Infrastructure (V2I) communication in current transportation systems, Traffic Light Detection (TLD) is still considered an important module in autonomous vehicles and Driver Assistance Systems (DAS). To overcome low flexibility and accuracy of vision-based heuristic algorithms and high power consumption of deep learning-based methods, we propose a lightweight and real-time traffic light detector for the autonomous vehicle platform. Our model consists of a heuristic candidate region selection module to identify all possible traffic lights, and a lightweight Convolution Neural Network (CNN) classifier to classify the results obtained. Offline simulations on the GPU server with the collected dataset and several public datasets show that our model achieves higher average accuracy and less time consumption. By integrating our detector module on NVidia Jetson TX1/TX2, we conduct on-road tests on two full-scale self-driving vehicle platforms (a car and a bus) in normal traffic conditions. Our model can achieve an average detection accuracy of 99.3 percent (mRttld) and 99.7 percent (Rttld) at 10Hz on TX1 and TX2, respectively. The on-road tests also show that our traffic light detection module can achieve $ ± 1 . 5 m errors at stop lines when working with other self-driving modules.

Machine learning

Abstract: Machine Learning is the study of methods for programming computers to learn. Computers are applied to a wide range of tasks, and for most of these it is relatively easy for programmers to design and implement the necessary software. However, there are many tasks for which this is difficult or impossible. These can be divided into four general categories. First, there are problems for which there exist no human experts. For example, in modern automated manufacturing facilities, there is a need to predict machine failures before they occur by analyzing sensor readings. Because the machines are new, there are no human experts who can be interviewed by a programmer to provide the knowledge necessary to build a computer system. A machine learning system can study recorded data and subsequent machine failures and learn prediction rules. Second, there are problems where human experts exist, but where they are unable to explain their expertise. This is the case in many perceptual tasks, such as speech recognition, hand-writing recognition, and natural language understanding. Virtually all humans exhibit expert-level abilities on these tasks, but none of them can describe the detailed steps that they follow as they perform them. Fortunately, humans can provide machines with examples of the inputs and correct outputs for these tasks, so machine learning algorithms can learn to map the inputs to the outputs. Third, there are problems where phenomena are changing rapidly. In finance, for example, people would like to predict the future behavior of the stock market, of consumer purchases, or of exchange rates. These behaviors change frequently, so that even if a programmer could construct a good predictive computer program, it would need to be rewritten frequently. A learning program can relieve the programmer of this burden by constantly modifying and tuning a set of learned prediction rules. Fourth, there are applications that need to be customized for each computer user separately. Consider, for example, a program to filter unwanted electronic mail messages. Different users will need different filters. It is unreasonable to expect each user to program his or her own rules, and it is infeasible to provide every user with a software engineer to keep the rules up-to-date. A machine learning system can learn which mail messages the user rejects and maintain the filtering rules automatically. Machine learning addresses many of the same research questions as the fields of statistics, data mining, and psychology, but with differences of emphasis. Statistics focuses on understanding the phenomena that have generated the data, often with the goal of testing different hypotheses about those phenomena. Data mining seeks to find patterns in the data that are understandable by people. Psychological studies of human learning aspire to understand the mechanisms underlying the various learning behaviors exhibited by people (concept learning, skill acquisition, strategy change, etc.).

Pattern Recognition and Machine Learning

Abstract: Probability Distributions.- Linear Models for Regression.- Linear Models for Classification.- Neural Networks.- Kernel Methods.- Sparse Kernel Machines.- Graphical Models.- Mixture Models and EM.- Approximate Inference.- Sampling Methods.- Continuous Latent Variables.- Sequential Data.- Combining Models.