Ramesh Nadavati — Research

1. Project: Your Face Tells Everything: An Analysis of Human Emotions with Internet and without Internet | Guide:Prof. Dr. Charlotte Tang

Objective: To analyze the user emotion and also the usage of emoticons in daily life.

Outcome: From this study, I tried to analysis and provide the proof of usage of emoticons according to the age factor. To compute the results, I have collected the data from 63 users around the globe. The final results were intermediate, but 3rd age group people were using the most emoticons for different things/meanings.

2. Project: The role of machine learning concepts in biometric recognition of face and iris | Guide:Prof. Dr. Michael Farmer

Objective: The main goal of machine learning concept is to reach a state when machines are able to respond without humans explicitly programming them. Machine learning concepts are efficiently used for feature extraction and classification and also directly applicable to biometric systems.This paper purely concentrating on feature extraction and classification.

Outcome: Worked on machine learning concepts related to Biometric recognition, for this project reviewed and discussed more than 30 articles and understood some important concepts. The final report was submitted, including four supervised Machine Learning algorithms like ‘support vector machine’, ’Fuzzy expert system’, ‘Gaussian mixture model’, ‘Artificial Neural Network’ in biometrics research. finally tested the dataset and have noted as much as 73% of accuracy.

3. Bachelor’s Thesis: Dc-Distribution Application | Guide:Prof. M.Purushotham

Overview: In a dc-distribution system, a bidirectional inverter is required to control the power flow between dc bus and ac grid, and to regulate the dc bus to a certain range of voltages. A droop regulation mechanism is required, according to the inverter inductor current levels to reduce capacitor size, balance power flow, and accommodate load variation is proposed. Since the PV array voltage can vary from 0to600v, especially with thin-film PV panels, the MPPT topology is formed with buck and boost converters to operate at the dc-bus voltage around 380v.