Cases of theft and robbery of computers, CCTV equipment, and LCD projector have become more frequent in schools. In addition, fire hazards are great threat to educational institutions where expensive learning materials are kept. Such incidents could be lessened and avoided if schools are equipped with appropriate security systems capable of monitoring and informing people about the coming possible danger. Thus, the development of Fire and Motion Early Warning Device (FMEWD) is timely and relevant. FMEWD consists of a website and interconnected devices and sensors intended to provide an efficient and effective warning system for preventing incidents relating to fire, smoke, and intrusion within an office. Upon detection, the system automatically sends an email and SMS to registered users. This study used the Agile Development Model which allows features to be delivered quickly and more frequently with higher levels of predictability. Evidently, the integration of different technologies conceptualized by the researcher addresses the pressing security concerns faced by educational institutions like NEUST.

Vehicle license number plate detection is essential for road safety and traffic management. Many existing systems have been proposed to achieve high detection precision without optimization of computer resources. Existing models have not preferred to use devices like smartphones or surveillance cameras because of high latency, data loss, bandwidth costs, and privacy. In this article, we propose a model of unloading decisions based on reinforcement learning (RL-OLD) for recognition and detection of vehicle license plates for high precision with optimization of computer resources. The proposed model detected different categories of vehicle registration plates by effectively utilizing edge computing. Our model can choose either the compute-intensive model of the cloud or the lightweight model of the local system based on the properties of the number plate. This approach has achieved high accuracy, limited data loss, and limited latency.

The important need for suitable indoor positioning systems has recently seen an exponential rise with location-based services emerging in many sectors of human life. This has led to adopting techniques to mine location data to discover useful insights to improve the accuracy of the various indoor positioning systems. Although indoor positioning has been reviewed in some literary works, an in-depth survey of how data mining could improve the performance of indoor localization systems is still lacking. This paper surveys data mining techniques such as Na¨ıve Bayes, Regression, K-Means, K-Nearest Neighbor (KNN), Support Vector Machines (SVM), Random Forest (RF), Expectation Maximization (EM), Neural Networks (NN), and Deep Learning (DL) including how they were used to improve the accuracy of indoor positing systems using various supporting technologies such as WiFi, Bluetooth, Radio Frequency Identification (RFID), Visible Light Communication (VLC), and indoor localization techniques such as Received Signal Strength Index (RSSI), Channel State Information (CSI), fingerprinting, and Time of Flight (ToF). Additionally, we present some of the challenges of existing indoor positioning systems that employ data mining while highlighting areas of future research that could be exploited in addressing those challenges.

In the manufacturing, process a burr has been observed during the drilling through a hole in an aluminum bar. From the view of the life of a product, minimization of the burr should be significant. So in this research main aim is to identify how input parameters: drill diameter, point angle & spindle speed influenced output parameters burr height & thickness. To execute this operation a total of 27 examinations on an Aluminum 6061T6 plate is taken. Overall research performed into two stages. In first stage, Surface response methodology is used to design two objective functions for burr height & thickness with the help of input parameters and then these two objective functions combined to construct a single objective function. In next stage improved version of elephant swarm optimization (ESWSA) algorithm is applied to get the optimum input parameters. The predicted output variable after the optimization techniques (Test 2 & Test 3) further checked with experimental result to determine the accuracy of the proposed model. In a conclusion section it is seen that the average error of drill diameter, drill point angle & spindle speed are 1.72%, 3.84% & 3.89% respectively with average RMSE is 2.56 *10^-6. For further validation of effectiveness of proposed model is also compared with the state of art techniques in the field burr minimization.

Soil mechanics problems deal with various types of soil that exhibit erratic behaviour in the real world, one such soil being the expansive soil where it takes a lot of laboratory test procedures to ascertain the physical properties of this soil. Modeling the behaviour of the expansive soil is complex and sometimes beyond the aptitude of most traditional procedures of physically-based engineering approaches. Artificial neural networks (ANN) are the ones used for predicting the complex nature of the soil since it has shown superior predictive potential as compared to the conventional approaches. This review aims to deliver and discuss the numerous applications of artificial neural network technique accomplished by various researchers in the field of geotechnical engineering to predict several properties of the expansive soil such as free swell index, unconfined compressive strength, shear strength of the soil, swelling pressure and swell percent, compaction characteristics, and plasticity index. This paper will assist practising engineers in determining the best modelling approaches and formulating the necessary data for using the ANN technique to solve soil mechanics problems.