Soham Mukherjee

Graduate Research Assisstant

Purdue University

Biography

Hii, I graduated from Purdue University with a Ph.D. in Computer Science. My research focus was/is the synergy of Topological Data Analysis (TDA) and Deep Learning, with an emphasis on Graph Neural Network (GNN). My thesis, “Unveiling patterns in data:harnessing computational topology in machine learning” is publicly available here. I moved to Purdue (Fall 2020) from The Ohio State University along with my advisor Prof. Tamal Krishna Dey. I am in CTDA group of Purdue CS dept. I completed my Bachelors in Electronics and Tele-communication Engineering from Jadavpur University, Kolkata. My undergraduate research was in Circuits and Systems in ADESL lab under Prof. Mrinal Kanti Naskar. I am from the city of joy, Kolkata, situated in eastern part of India. Apart from doing nerdy stuff(Nerds call it research),I sometimes pick up my good old camera to click some photos.

Interests

Topological Data Analysis (TDA)
Applications of TDA in ML & Deep Learning
ML Applications

Education

PhD Student, 2020 - 2024

Purdue University, West Lafayette, IN
PhD Student, 2017 - 2020

The Ohio State University
BETCE, 2013 - 2017

Jadavpur University

Experience

Graduate Research Assistant

Purdue University

Aug 2020 – Present West Lafayette, IN

Responsibilities include:

Research
Developing Algorithms

Research Intern

May 2022 – Aug 2022 Yorktown Heights, NY

Research Intern

Engineering Intern

May 2021 – Aug 2021 Columbus, OH

Engineering Intern

Graduate Research Assistant

The Ohio State University

Aug 2017 – Aug 2020 Columbus, OH

Moved to Purdue with my advisor.

Undergraduate Researcher

Jadavpur University

Aug 2013 – Jul 2017 Calcutta, India

FPGA implementation of stochastics circuits under Dr. Mrinal Kanti Naskar

Accomplishments

IBM Research intern

IBM Research May 2022 – Aug 2022

Summer intern in Physna inc.

Physna inc. May 2021 – Aug 2021

DAAD Summer Research Intern

Georg-August Universität Göttingen May 2016 – Aug 2016

Evaluation of Waspmote Cryptography

See certificate

Projects

A Jacobi-set based loss function for segmentation task

Segmentation of fine-scale structures in natural and bio-medical images are gaining importance with the development of high resolution electron microscopy images. The task still remains challenging as per-pixel accuracy is not only the metric of concern because of the imbalance in the dataset. In this project, a new loss function based on the Jacobi-sets are proposed.

Soham Mukherjee

Graph generation with Geometrical and Topological Constraints

Persistent homology, a tool from computational topology we computed persistent diagrams of graphs and incorporated topological & geometrical constraints while generating graphs.

GEFL: Extended Filtration Learning for Graph Classification

Learning Extended filtration on graphs

Simon Zhang, Soham Mukherjee, Tamal K. Dey

Predicting phenotypes using topology and machine learning

TDA helps to identify phenotypes.

Sayan Mandal, Soham Mukherjee

Determining clinically relevant features in cytometry data using persistent homology

TDA helps to identify structural difference in single-cell cytometry data.

Featured Publications

GEFL: Extended Filtration Learning for Graph Classification

Extended persistence is a technique from topological data analysis to obtain global multiscale topological information from a graph. This includes information about connected components and cycles that are captured by the so-called persistence barcodes. We introduce extended persistence into a supervised learning framework for graph classification. Global topological information, in the form of a barcode with four different types of bars and their explicit cycle representatives, is combined into the model by the readout function which is computed by extended persistence. The entire model is end-to-end differentiable

Simon Zhang, Soham Mukherjee, Tamal K. Dey

Denoising with discrete Morse theory

Denoising noisy datasets is a crucial task in this data-driven world. In this paper, we develop a persistence-guided discrete Morse …

Soham Mukherjee

Determining clinically relevant features in cytometry data using persistent homology

Identifying differences between cytometry data seen as a point cloud can be complicated by random variations in data collection and data sources. We apply persistent homology used in topological data analysis to describe the shape and structure of the data representing immune cells in healthy donors and COVID-19 patients. By looking at how the shape and structure differ between healthy donors and COVID-19 patients, we are able to definitively conclude how these groups differ despite random variations in the data. Furthermore, these results are novel in their ability to capture shape and structure of cytometry data, something not described by other analyses.

Soham Muhkerjee, Darren Wethington, Tamal K. Dey, Jayajit Das

Gene expression data classification using topology and machine learning models

We show that the representative cycles we compute have an unsupervised inclination towards phenotype labels. This work thus shows that topological signatures are able to comprehend gene expression levels and classify cohorts accordingly.

Soham Mukherjee, Sayan Mandal, Tamal K. Dey

Recent Publications

GEFL: Extended Filtration Learning for Graph Classification

Simon Zhang, Soham Mukherjee, Tamal K. Dey

Courses

Purdue University

CS59900 - Machine Learning on Graphs
CS59900 - Topological Data Analysis
CS57700 - Natural Language Processing

Ohio State University

CSE 5523 - Statistical Pattern Recognition and Machine Learning
CSE 5525 - Deep Learning
CSE 6331 - Algorithms
CSE 6332 - Advanced Algorithms
CSE 5542 - Real-time Rendering
CSE 5543 - Geometric Modeling
MATH 6801 - Algebraic Topology 1
MATH 4556 - Dynamical Systems

Contact