International Journal of Intelligent Systems Technologies and Applications (7 papers in press)
Clustering based hybrid resampling techniques for social lending data
by Pankaj Kumar Jadwal, Sonal Jain, Basant Agarwal
Abstract: Social lending is the most popular and emerging loan disbursement process where an individual can act as a borrower or lender. Credit risk evaluation of the borrowers in an effective way is a crucial task, especially in social lending, where chances of being defaulted are more than the traditional models. Social lending datasets are imbalanced in nature due to the low number of defaulters than successful borrowers. Machine learning models based on such datasets contain biasing towards the class representing the majority of samples (Majority class). Therefore, the probability of accurate prediction of minority class samples is decreased due to biasing towards majority class samples. In this paper, we propose a novel Clustering based Hybrid Sampling algorithm (CBHS), where multi-phase K-means clustering is applied on the minority class samples to perform oversampling (KMBOS), and Fuzzy c -means clustering is used on the majority class samples to perform undersampling (FCBU). Experiments results show that KMBOS and FCBU algorithms outperform state of the art techniques of oversampling and undersampling.
Keywords: Credit risk; Clustering; Classification; Hybrid model; Oversampling; Undersampling; Class Imbalance.
ADAPTIVE FUZZY LOGIC CONTROL FOR A ROBOTIC GAIT TRAINING ORTHOSIS
by Deepa Mathur, Deepak Bhatia, Prashant K. Jamwal, Shahid Hussain, Mergen H. Ghayesh
Abstract: This paper aims to develop an adaptive control strategy for a fuzzy logic system to be implemented in a robotic gait training orthosis. The robotic orthosis has a bio-inspired design which has evolved after a careful study of the biomechanics of human gait. Ambulatory requirements of the robot have been achieved by employing light weight but powerful pneumatic muscle actuators (PMA). The sagittal plane rotations achieved by the robotic orthosis at the hip and knee are achieved by implementing a Pneumatic Muscle Actuator (PMA) for actuation. The PMA of the Robotic orthosis was controlled by a fuzzy logic controller based on the Mamdani inference in order to obtain the necessary rotational degrees of freedom. To cope with the nonlinear behavior of PMA towards external disturbances, a second instance of fuzzy based controller has been developed. The PMA is infamous for its time dependent characteristics hence an adaptive control mechanism has been introduced in an attempt to compensate for it. Healthy subjects were employed for performing experiments in order to understand and estimate the performance of the adaptive fuzzy logic controller as well as the entire adaptive robotic design. The human-robot interaction was mainly maintained passive-active, while the paths used for the robot were strictly predefined trajectories which were usually employed by physical therapists during rehabilitation sessions.
Keywords: Adaptive fuzzy logic control; robotic orthosis; gait training; PMA; neurological impairments.
A reliability-aware scheduling algorithm for parallel task executing on cloud computing system
by Jie Cao, Zhifeng Zhang, Bo Wang, Xiao Cui, Jinchao Xu
Abstract: As cloud computing is established on the massive cheap server clusters, which causes compute nodes software and hardware to go wrong. Different computing nodes and communications links have different failure rate. For the parallel task scheduling problem that cloud users have requirements for deadlines and executing reliability, we put forward to generate all possible execution schemes of a parallel task on a cloud computing system. All the execution schemes are constructed into an execution scheme graph (ESG), in which a path from the start point to end point corresponds to an execution scheme of a parallel task. Based on ESG, we propose the maximum reliability execution scheme solving algorithm MRES that searches the execution schemes which have maximum reliability cost while meeting the parallel tasks deadline requirement. The experimental results show that MRES algorithm can effectively improve the executing success rate.
Keywords: Cloud computing;Reliability;Directed acyclic graph;Task scheduling.
Innovative and Affordable Wearable Solution for Suppression of Hand Tremors
by Shriram K Vasudevan, Dana Vishnu, G.A. Dhanush, Kiran S Raj, S. Siddharth
Abstract: Hand tremor influences a huge number of individuals around the world. Tremor is characterized as unintended, cadenced muscle development. Albeit most tremors neither reason serious medical issues nor threaten life, most patients experience many troubles in performing everyday exercises. It is humiliating to numerous individuals, particularly more seasoned people. Tremor negatively affects personal satisfaction and decreases the ability to perform everyday life undertakings such as eating, drinking, writing, and changing clothing. Some patients with mild tremor do not visit doctors if tremor does not hinder their daily life activities. Their situations often worsen over time. There is currently very few to almost no external solution existing to cope with this medical condition. We have developed a glove that is able to dynamically conceive the motion of a users hand and ascertain the difference between intentional movement and unintentional movement. Further, we have quantified, verified and attempted to provide a haptic response to the users fingers with the help of coin motors to help counteract the unwanted vibration.
Keywords: Parkinson’s Disease; Medical IoT; AI; Sensors for medicine;rnElectroactive Polymers; Hand Tremors; AI for medicine.
Adaptive vision-based system for landing an autonomous hexacopter drone on a specific landing platform
by Abdel Ilah Alshbatat
Abstract: Today, autonomous flight and the precise landing of drones are crucial for many applications such as object detection, delivery services, and the automated battery charging process. In all of these applications, automated drone missions will greatly improve its capabilities and even allow a broader range of applications. Such a process required a safe autonomous flight and a precise landing on the designated landing platform. The focus of this research is to present an adaptive vision-based system for landing an autonomous Hexacopter Unmanned Aerial Vehicle (UAV) on a specific landing platform. The system is based on a color-based algorithm to position a visual markers and then land the drone in environments where GPS signal is weak or unavailable. The prototype is divided into three subsystems: landing platform, imaging sensors with a Raspberry Pi as a companion computer, and the Hexacopter drone. Four cameras are used in the suggested approach. One camera is connected directly to the flight controller and the other three are connected to the companion computer. The landing platform consists of special visual markers made up of four circles with different sizes, and different colors. The Hexacopter frame is equipped with a Pixhawk flight controller and six motors. The attitude and altitude of the drone are adjusted automatically based on the information obtained from the whole vision system. For any uncertainty in the information received from the landing vision system, an adaptive vision-based algorithm is used. To verify the performance of the proposed system, a practical test bench based on the designed and fabricated prototype was developed. Simulation studies and real experiments were conducted under different scenarios. The results show that the proposed system is highly reliable and robust in detecting the landing area and landing the drone on the designed markers.
Keywords: Hexacopter; Unmanned Aerial Vehicle (UAV); autonomous landing; vision-based system; color-based object detection.
Real Time Vision Based Controller for Delta Robots
by Ali Sharida, Iyad Hashlamon
Abstract: This paper investigates two real time vision based control algorithms for delta robots. The first one aims to enable the robot to track different objects based on their colours and shapes. This algorithm does not need any initial calibration, instead, it depends on Least Square Algorithm (LSA) to generate the required transformation matrices, also it is implemented on a standalone controller with no additional time complexity to the main controller. The second one is a self-calibrating human hand gestures tracking algorithm, which can automatically calibrate itself and generate transformation matrices automatically. The two algorithms were designed, implemented and scheduled in real time manner. The results show that these algorithms can track fast moving objects effectively regardless of the initial configuration of the robot. These algorithms provide important solutions for common problems related to visual servoing such as field of view and calibration.
Keywords: Vision based control; Real Time Control; Delta Robot; Visual servoing; hand gestures tracking.
Blockchain for Smart Grid Security: Applications, Trends, and Challenges
by Diogo Mattos, Dianne Medeiros, Diego Passos, Natalia Fernandes, Débora Muchaluat-Saade, Igor Moraes, Celio Alburquerque
Abstract: The electric power grid is the world's largest engineering system, and its secure and reliable operation is vital to human activities. The introduction of intelligence in the electrical power grid through smart grids imposes challenges that require new techniques and approaches to provide cyber-physical security. In this article, we discuss the use of blockchain to provide security and reliability to smart grids. Blockchain allows untrusted nodes to correctly and verifiably interact with each other in a distributed peer-to-peer network without any reliable intermediary. We explore smart contracts, codes resident in blockchain that automate multi-step processes to trade electric energy automatically. We also discuss initiatives, challenges, and research opportunities of blockchain technologies in the electrical sector.
Keywords: Blockchain, Smart Grid, Security, Power Grid, Network Communication