Data Mining - Concepts and Techniques.

Activity detection and classification are very important for autonomous monitoring of humans for applications, including assistive living, rehabilitation, and surveillance. Wearable sensors have found wide-spread use in recent years due to their ever-decreasing cost, ease of deployment and use, and ability to provide continuous monitoring as opposed to sensors installed at fixed locations. Since many smart phones are now equipped with a variety of sensors, such as accelerometer, gyroscope, and camera, it has become more feasible to develop activity monitoring algorithms employing one or more of these sensors with increased accessibility. We provide a complete and comprehensive survey on activity classification with wearable sensors, covering a variety of sensing modalities, including accelerometer, gyroscope, pressure sensors, and camera- and depth-based systems. We discuss differences in activity types tackled by this breadth of sensing modalities. For example, accelerometer, gyroscope, and magnetometer systems have a history of addressing whole body motion or global type activities, whereas camera systems provide the context necessary to classify local interactions, or interactions of individuals with objects. We also found that these single sensing modalities laid the foundation for hybrid works that tackle a mix of global and local interaction-type activities. In addition to the type of sensors and type of activities classified, we provide details on each wearable system that include on-body sensor location, employed learning approach, and extent of experimental setup. We further discuss where the processing is performed, i.e., local versus remote processing, for different systems. This is one of the first surveys to provide such breadth of coverage across different wearable sensor systems for activity classification.

A Survey on Activity Detection and Classification Using Wearable Sensors

Data fusion and multiple classifier systems for human activity detection and health monitoring: Review and open research directions

ABSTRACT The accurate diagnosis of the initial stage COVID-19 is necessary for minimizing its spreading rate. The physicians most often recommend RT-PCR tests; this is invasive, time-consuming, and ineffective in reducing the spread rate of COVID-19. However, this can be minimized by using noninvasive and fast machine learning methods trained either on labeled patients’ symptoms or medical images. The machine learning methods trained on labeled patients’ symptoms cannot differentiate between different types of pneumonias like COVID-19, viral pneumonia, and bacterial pneumonia because of similar symptoms, i.e., cough, fever, headache, sore throat, and shortness of breath. The machine learning methods trained on labeled patients’ medical images have the potential to overcome the limitation of the symptom-based method; however, these methods are incapable of detecting COVID-19 in the initial stage because the infection of COVID-19 takes 3 to 12 days to appear. This research proposes a COVID-19 detection system with the potential to detect COVID-19 in the initial stage by employing deep learning models over patients’ symptoms and chest X-Ray images. The proposed system obtained average accuracy 78.88%, specificity 94%, and sensitivity 77% on a testing dataset containing 800 patients’ X-Ray images and 800 patients’ symptoms, better than existing COVID-19 detection methods.

https://www.tandfonline.com/doi/pdf/10.1080/08839514.2022.2055398?needAccess=true

Initial Stage COVID-19 Detection System Based on Patients’ Symptoms and Chest X-Ray Images

Ambient Assisted Living using mobile device sensors is an active area of research in pervasive computing. Multiple approaches have shown that wearable sensors perform very well and distinguish falls reliably from Activities of Daily Living. However, these systems are tested in a controlled environment and are optimized for a given set of sensor types, sensor positions, and subjects. In this work, we propose a self-adaptive pervasive fall detection approach that is robust to the heterogeneity of real life situations. Therefore, we combine sensor data of four publicly available datasets, covering about 100 subjects, 5 devices, and 3 sensor placements. In a comprehensive evaluation, we show that our system is not only robust regarding the different dimensions of heterogeneity, but also adapts autonomously to spontaneous changes in the sensor’s position at runtime.

Hips Do Lie! A Position-Aware Mobile Fall Detection System

Design of a Smart Safety Device for Women using IoT

This paper aims to develop a classification system to distinguish COVID-19 positive and negative cases based on common symptoms and could be used as a first-level screening tool for early detection of mild cases. Accordingly, existing classification models such as Logistic Regression, Gradient Boosting (GB), Random Forest (RF) and K-Nearest Neighbours have been tried on the COVID-19 symptoms dataset to identify the best performing model. Although traditional machine learning models provide promising results in terms of accuracy, precision and recall, this paper analyses the possibilities of improvement in classification results through ensemble and hybrid approaches. It is observed from the results that K-mode clustering followed by classification-based hybrid modelling resulted in improved classification accuracy in the clusters leading to an average accuracy of 87.17% and 87.24% with GB and RF respectively. Finally, the MaxVoting ensemble model, comprising GB and RF algorithms further boosted the accuracy to almost 90%.

/pdf/symptoms-based-early-clinical-diagnosis-of-covid-19-cases-4s30bkviqu.pdf

Symptoms based Early Clinical Diagnosis of COVID-19 Cases using Hybrid and Ensemble Machine Learning Techniques

In face recognition, searching and retrieval of relevant images from a large database form a major task. Recognition time is greatly related to the dimensionality of the original data and the number of training samples. This demands the selection of discriminant features that produce similar results as the entire set and a reduced search space. To address this issue, a Multi-Level Search Space Reduction framework for large scale face image database is proposed. The proposed approach identifies discriminating features and groups face images sharing similar properties using feature-weighted Fuzzy C-Means approach. A hierarchical tree model is then constructed inside every cluster based on the discriminating features which enables a branch based selection, thereby reducing the search space. The proposed framework is tested on three benchmark and two self-created databases. The experimental results show that the proposed method achieved an average accuracy of 93% and an average search time reduction of 66% compared to existing approaches for search space reduction of face recognition.

Multi-Level Search Space Reduction Framework for Face Image Database

Human activity recognition (HAR) has gained a lot of significance in monitoring the health of people, especially to detect fall among elderly people who live independently. This project proposes a novel method for recognizing activities and detecting fall of a person using body-worn sensors. Traditional algorithms like Naive Bayes classifier and Support Vector Machine are mainly used for activity classification. However, these systems fail to capture significant association that exists between interesting patterns. Existing accelerometer based wearable systems are not sufficient to determine the fall of a person. Hence, a Fuzzy Associative Classification based Human Activity Recognition (FAC-HAR) system is proposed to overcome the aforementioned drawbacks in detecting abnormal status of a person. The proposed (FAC-HAR) system uses three different sensors namely heartbeat, breathing rate and accelerometer and employs fuzzy clustering and associative classification for abnormality detection. The proposed system introduces a novel learning mechanism is to improve classification accuracy. A classification accuracy of 92% is achieved with the proposed fuzzy frequent pattern mining based human activity recognition.

Adaptive learning based human activity and fall detection using fuzzy frequent pattern mining

Mining data over streams in one pass and using constant memory is a challenging task. Decision trees are one of the most popular classifiers for both batch and incremental learning due to their high degree of interpretability, ease of construction and good accuracy. The most popular decision tree for stream classification is Hoeffding Tree based on Hoeffding bound. Literature shows a few variants of decision trees based on different bounds. The default class prediction method adopted in decision tree is “majority class” approach. Later, the accuracy of prediction was scaled up by a hybrid decision tree where Naive Bayes classifier was used for prediction. Kernel Density Estimation (KDE) is employed in Flexible Naive Bayes for classification. However, it is suitable for modeling static data set. This paper proposes an Incremental Flexible Naive Bayes (IFNB) based hybrid decision tree paradigm that uses KDE to model continuous attributes at leaf nodes of the tree for improving the class prediction accuracy. Experimental results on both synthetic and real dataset show that the proposed IFNB based leaf classifiers achieves improvement over the class prediction methods adopted in existing decision trees for data streams.

C. Sweetlin Hemalatha

Papers

Design of a Smart Safety Device for Women using IoT

Symptoms based Early Clinical Diagnosis of COVID-19 Cases using Hybrid and Ensemble Machine Learning Techniques

Multi-Level Search Space Reduction Framework for Face Image Database

Adaptive learning based human activity and fall detection using fuzzy frequent pattern mining

Hybrid decision trees for data streams based on Incremental Flexible Naive Bayes prediction at leaf nodes