Description

Guide: Dr. Kakoli Banerjee (Associate Professor, Department of Computer Science, JSSATEN)

  • Water Quality Index Computation & Estimation Using Machine Learning Techniques

Submitted to MDPI Water for review

A Research Project which estimates and analyses the Water Quality Index in Uttar Pradesh, India. It finds a subset of other known physiochemical water quality parameters affecting WQI the most from a collected dataset for its prediction using Supervised Machine Learning algorithms.

This project received a grant of INR 300,000 from the Collaborative Research and Innovation Program.

(1) Dataset of 51 collected groundwater samples from Gautam Buddha Nagar

(2) Computes WQI for groundwater samples using the Brown et al. method

(3) Employs Supervised Machine Learning algorithms (Linear Regression, Support Vector Regressor, Decision Tree and Random Forest Regressor) for the prediction of WQI

(4) Reduces the dimensionality of computation of WQI to a single variable i.e. Turbidity

  • Accessibility System Using Machine Learning

We first analysed recent implementations of Sign Language Recognition using Machine Learning Techniques and Artificial Intelligence by outlining the methodology used in the steps of Data Acquisition, Feature Extraction and Model Training. The review paper published and presented discusses all the gaps in the existing studies along with the various tools used for the implementation. Furthermore, we presented an accessibility system to help differently-abled people access and interact with computers. We illustrated the creation of a novel JavaScript-based text-entry mechanism for braille users that uses position keys on regular keyboards to “position” the user’s six fingers and form a 6-dot braille keyboard and also discussed the development of an LSTM-based Sign-to-Text model created on a custom dataset of American Sign Language.

(1) Review of Existing Artificial Intelligence based Sign Language Recognition Techniques – Review paper published on 🔗 IEEE Xplore (2) Full implementation paper accepted to be published in Springer’s SN Computer Science