Contact Information
karan.r.grewal@gmail.com
Google Scholar
LinkedIn

About Me

Hi there! I'm a Machine Learning Researcher & Engineer based in San Francisco. At present, I develop driver assistance technologies at Helm.ai. Prior to this, I worked on bringing ideas from neuroscience into the world of computer algorithms at Numenta.

Check out my recent episode on the Data Science at Home podcast where I discuss neuroscience, computer algorithms, and their interplay. (Also available on Spotify and Apple Podcasts.)

Published Work

  • Avoiding Catastrophe: Active Dendrites Enable Multi-Task Learning in Dynamic Environments
    Abhiram Iyer, Karan Grewal, Akash Velu, Lucas O. Souza, Jeremy Forest, Subutai Ahmad.
    Frontiers in Neurorobotics, 2022.
    [paper]

  • Exploring the Structure of Human Adjective Representations
    Karan Grewal, Joshua C. Peterson, Bill Thompson, Thomas L. Griffiths.
    NeurIPS 2021 workshop on Shared Visual Representations in Human & Machine Intelligence.
    [paper]

  • Chaining Algorithms and Historical Adjective Extension
    Karan Grewal, Yang Xu.
    Computational Approaches to Semantic Change, 2021.
    [paper]

  • Chaining and Historical Adjective Extension
    Karan Grewal, Yang Xu.
    CogSci 2020.
    [paper]

  • Learning Deep Representations by Mutual Information Estimation and Maximization
    R Devon Hjelm, Alex Fedorov, Samuel Lavoie, Karan Grewal, Phil Bachman, Adam Trischler, Yoshua Bengio.
    ICLR 2019.
    [paper]

  • Variance Regularizing Adversarial Learning
    Karan Grewal, R Devon Hjelm, Yoshua Bengio.
    ICML 2017 workshop on Implicit Generative Models.
    [paper]


Outreach

alt text




The MISE Foundation aims to expose bright high school students in Ghana to the latest developments in STEM fields. At one of their recent annual summer programs, I served as a research mentor and led a select group of students through a project on designing algorithms to generate human face sketches. We also had long discussions about the ethics of such technology as its uses are highly controversial today. I hope my efforts will motivate these students to one day become leaders in these fields and transform the global R&D landscape.