A. Sheryl Oliver ¹, N. Maheswari ¹

Affiliations
1 School of Computing Science and Engineering, VIT University, Chennai- 632014, Tamil Nadu, IN

Abstract

Objectives: The Objective of this research is to develop a feature extractor and a classifier which will identify and classify the gestures of infants, elderly and pregnant woman using Gait Signal re-ceived from wearable electrodes which is positioned on the body of subjects. Methods/Statistical Analysis: Remote health care monitoring is a technology which enables monitoring a person outside usual medical settings i.e., in the house or residence, which may increase access to caretakers or person at home but it will decrease healthcare deliverance costs. Findings: A novel segmented pigeon hole data extraction and reduction technique is proposed for reducing data and feature extraction. Secondly an Iteration Reduced Threshold based Classifier (IR-Threshold Classifier) has been introduced, which classifies the reduced extracted data into Safe and Danger for toddler, Normal and contra for pregnant women and Stable and Fall for elderly. Feature extraction and reduction using Segmented Pigeon Hole algorithm reduced the dataset for this domain. It is compared with bench mark data set and it had produced the significant data reduction. The IR Threshold classifier had shown 95% of accuracy when compared with the other classifiers. Applications/Improvements: This gives the best predominant electrode set by reducing data which will increase the classification accuracy.

Keywords

Fall and Normal Category, Feature Extraction, IR-Threshold Classifier, Machine Learning Algorithm, Segmented Pigeon Hole, Wearable Electrodes.

Full Text

   

Nayan Mattani ¹, J. Sharath Kumar ¹, A. Prabakaran ¹, N. Maheswari ¹

Affiliations
1 School of computing science and Engineering, VIT University, Chennai - 600127, IN

Abstract

Objectives: This paper focuses on a privacy preservation technique which is applied on graphs to preserve sensitive information present as shortest paths. Methods/Statistical Analysis: Provide privacy in Social Network Analysis by protecting the sensitive edge weights with the help of preserving the nearest shortest path lengths as well as shortest paths so that individual confidential information can be protected from multiple type of attacks. This research work provides more privacy then greedy perturbation technique in social network analysis. Findings: Privacy preservation has a tradeoff between the utility of data and preservation of sensitive information. This is achieved by modification of shortest path length in graphs and also maintains the structure of the graph. This procedure enhances the privacy of sensitive information with minimal concerns to utility. Application/Improvements: The privacy preservation technique of edge weight perturbation is applied to social graphs in a small user group to preserve sensitive information when data is shared within the group members. The edge weight perturbation algorithm can be improved by combing the algorithm with the preservation techniques for the user nodes.

Keywords

Anonymization, Graphs, Perturbation, Privacy, Social Networks.

Full Text