Arka Prava Roy
About Arka Prava Roy
Arka Prava Roy is a Data Scientist with expertise in gaze data analysis and deep learning. He currently works at Intellithink Industrial IoT and has previously held positions at Deloitte and various research institutions.
Work at Intellithink Industrial IoT
Arka Prava Roy has been employed as a Data Scientist at Intellithink Industrial IoT since 2022. In this role, he focuses on building efficient models for the classification of biomedical signals. His work is conducted under the supervision of Dr. Hemanta Kumar Mondal, with the goal of implementing these models on a hardware platform.
Education and Expertise
Arka Prava Roy earned a Bachelor of Technology in Electronic and Communications Engineering Technology from the National Institute of Technology Durgapur, where he studied from 2017 to 2021. His education laid the foundation for his expertise in data science and deep learning, particularly in the analysis of gaze data and biomedical signals.
Background
Prior to his current position, Arka Prava Roy gained experience in various roles. He worked as an Analyst at Deloitte for one month in 2021 in Bengaluru, Karnataka, India. He also served as a Research Intern at the Indian Institute of Technology, Patna for three months in 2020, and as a Deep Learning Research Intern at the Indian Statistical Institute, Kolkata for two months in 2019.
Achievements
Arka Prava Roy has contributed to several significant projects in the field of data science. He utilized the mRMR algorithm to identify key features in gaze data, distinguishing between monolingual and bilingual reading. Additionally, he developed a multilabel classification model for the Kaggle Toxic Challenge Dataset and created a simple interface for hosting this trained model, accessible online.
Research Contributions
During his time at the Indian Statistical Institute, Kolkata, Arka Prava Roy developed deep neural networks for gaze data analysis, focusing on differentiating between native and second language reading using Bidirectional LSTM models. He also contributed to an attention layer-based deep neural network model for the EmotionGIF Challenge, achieving a notable ranking in the competition.