KARAN DHINGRA

About Me

Hi, I am a final year master's student at IIIT, Delhi and currently doing my internship at Ericsson Global AI Accelerator. I am also working on my thesis under the guidance of Dr. Pravesh Biyani on Scalable ETA prediction.

I wish to

delve into A.I. at a much deeper level, focusing on natural language understanding, computer vision, reinforcement learning, and being closely involved in advancing these areas.


Scalable ETA prediction

Thesis

This project aims to estimate the arrival time for a scalable SWARM system.We are using real-time GTFS data from the 1000 buses running in Delhi.

Ericsson GAIA

Internship

Joined the team under the guidance of Data Science Director, Dr. Sunil Kumar Vuppala. Currently working on XRL algorithms, an extension of XAI(explainable AI) for reinforcement learning.

2020

Phylogeny of tumor cells

Independent Study

Studied multiple states of the art techniques which are used to generate phylogeny from the tumor samples. Proposed a joint algorithm based on two states of the art techniques to improve its collective performance.

Automatic Colorization

Sampling Color Images from Grayscale and I.R.

Implemented an autoregressive state-of-the-art algorithm, which can generate multiple distinct, color ful images from a grayscale input, followed by an ImageGAN based architecture that translates I.R. images to RGB.

Context Understanding using GANs

CSE556: Natural Language Processing

Given two sentences, the aim is to understand the explicit context in logical relationships between sentences. • Implemented a GAN based approach for sentence generation conditioned on the logical relationship.

Quantum Circuit Decomposition

CSE622: Introduction to Quantum Computing

Studied multiple states of the art techniques which are used to generate phylogeny from the tumor samples. Proposed a joint algorithm based on two states of the art techniques to improve its collective performance.

Exploring and Analysing Quantum Probabilistic Models

CSE561: Probabilistic Graphical Models

Analysed how real-world experiments like Prisoner’s Dilemma violate classical probabilistic models. Explored quantum probabilistic models to accommodate these violations.

2019

Object Detection using CenterNet

CSE543: Machine Learning

Studied how object detection algorithms get affected by the sizes of objects, and their aspect ratio. Worked on the implementation of CenterNet, which aims to generalize object detection over different sizes by using heat-map to represent an object instead of a bounding box.

Smart Home Control using Reinforcement Learning

CSE643: Artificial Intelligence

Proposed a reinforcement learning algorithm to optimize energy consumption with a dual objective function of minimizing energy and maximizing user-preferences.
Published in IEEE SDMS Conference 2020.

Research Assistant

IIIT Delhi

Proposed a single channel source algorithm, which aims to synthesize clean speech from the mixture rather than noise removal.