Aparna Balagopalan

Aparna Balagopalan

University of Toronto

About me

I am a Research Software Engineer at WinterLight Labs. I graduated from the University of Toronto with a Masters degree in Applied Computing in January 2019. I graduated with honors from the Indian Institute of Technology, Guwahati in 2017. I am interested in research and applications of machine learning for healthcare.


Research Intern at Philips Innovation Campus, India

May 2017 - July 2017

Internship in the Research department applying machine learning to healthcare applications.Projects completed during internship include anonymization of personal health information in unstructured text and sequence modelling and prediction of patient data.

Visiting Scholar/DAAD WISE Awardee at TU Dresden, Germany

May 2016 - July 2016

Joint project between Computer Vision Lab and Computer Graphics Lab, TU Dresden. Project involved visualization of Convolutional Neural Networks (CNNs) at a layer-level by particle rendering and developing a gradient-based importance measure for each filter in a CNN.

Research Intern at IISc Bangalore

May 2015 - July 2015

Project involved implementation of various learning algorithms using Theano in Python.Also, entropy-based calculations using information encoded in model weights were carried out.


Programming languages:

C, C++, CUDA, Python, Java, Assembly (ARM)

Software Packages:

MATLAB, Multisim, Electric VLSI, Spice, R, Minitab

Deep Learning Frameworks :

Tensorflow , Theano , Caffe, Chainer, PyTorch


Using Label Word Embeddings in Image Classification

Machine Learning and Data Mining Course

Project on using label semantics for Image Classification to train models robust to adversarial examples. Rather than using cross-entropy loss, loss function using word embedding-based representation of labels was devised with significant results on Fashion MNIST and CIFAR-10 datasets using shallow networks.

Multithreading vs Python GIL: A study

Parallel Computer Architecture and Programming

Project involved studying the effects of the Python Global Interpreter Lock on libraries commonly used for machine learning (NumPy and Scikit-learn) by profiling for concurrency, locks, hotspots of time etc. An adaptive way to set GIL check interval was introduced as a result of profiling.