Deya Chatterjee



I am currently a CS undergraduate in my pre-final year at SRM Institute of Science and Technology, Chennai, India. My current academic interests are Machine Learning, Deep Learning and NLP, but I am also looking forward to exciting work in other domains such as biology or social science. I am interested in deploying novel AI models to solve real problems that affect the world. I am also an impassioned advocate of women in CS (and more generally, in STEM), and involved in several outreach activities to realize the purpose.

Currently I am a data scientist working on applying convolutional neural nets and capsule nets to leukemia detection at AI Without Borders, and a student researcher at SRM Deep Learning Research Lab, where we experiment on varied deep learning projects as a team. I have worked as a research intern at IIIT Bangalore under Prof. Shrisha Rao in Summer 2018, where I worked on computational sustainability. I've also completed a research internship on handwriting recognition at the Computer Vision and Pattern Recognition Unit in Indian Statistical Unit, Kolkata in Winter 2017.

Email  /  CV  /  Github  /  LinkedIn

Research

I'm interested in AI, machine learning, deep learning and NLP. Much of my research is about leveraging AI in healthcare, especially to cancer surveillance. I have also worked on computational sustainability, and diverse side projects and course projects.


Selected Projects and Recent Work

I have worked on domains ranging from machine learning to computational sustainability. These are a few of my selected projects. For more info, please check out my Github.

Lung Cancer Detection using Deep Learning

This project was done using convolutional neural networks and generative adversarial networks (DCGAN, in particular) to detect lung cancer from CT scans of lesions and nodules in the lung. Improvements to the project, e.g., experimenting with different architectures and models to compare the performance is currently ongoing.

Information Retrieval from Cancer Pathology Reports

This project explores clinical IR applied to pathology reports, mostly those pertaining to cancer. Traditional NLP techniques were applied to report classification, but also current DL techniques such as hierarchical attention networks were implemented to obtain good performance.

Computational Sustainability: A Socio-Technical Perspective

This was the topic of my work at my latest internship. We performed extensive research on computational sustainability and sustainable computing. Sustainability is a very relevant topic in present times, and we aimed to understand how computational techniques can help achieve the goal of sustainability. The result of this work is a research paper, submitted and under review in ACM CSUR.

Online Handwriting Recognition

This early project was based on online Bangla handwriting recognition in Handwriting recognition Lab at ISI. I applied machine learning techniques for the task of preprocessing of online handwriting recognition data, and also worked on image processing techniques using PIL for the same.




  • Data Structures & Algorithms
  • Object Oriented Programming in C++
  • Object Oriented Programming in Java
  • Program Design and Development with C
  • Digital System Design
  • Computer Organization & Architecture
  • Microprocessors and Microcontrollers
  • Algorithms - Design & Analysis
  • Calculus
  • Probability & Queuing Theory
  • Discrete Mathematics
  • Deep Learning Specialization
  • Neural Networks and Deep Learning
  • Improving Deep Neural Networks: Hyperparameter Tuning, Regularization and Optimization
  • Structuring Machine Learning Projects
  • Convolutional Neural Networks
  • Sequence Models
  • Machine Learning (Prof. Andrew Ng.)
  • Cryptography
  • Social and Economic Networks: Models and Analysis

Last Updated: September 5,2018
Template credits: Jon Barron