Publications and Projects

Space to Policy: Scalable Brick Kiln Detection and Automatic Compliance Monitoring with Geospatial Data

Object Detection

Deep Learning

Satellite Imagery

Paper accepted in ACM Journal on Computing and Sustainable Societies

Eye in the Sky Detection and Compliance Monitoring of Brick

Deep Learning

Object Detection

Paper accepted in ACM Compass (Conference on Computing and Sustainable Societies) 2024 Poster track.

Retrieval-Augmented Generation(RAG) and Hypothetical Document Embeddings~(HyDE)

NLP

Large Language Models (LLMs), such as those driving today’s most advanced AI chatbots and information retrieval systems, are incredibly powerful. However, they’re not…

Image-to-Image for Climate Modelling using Auto-Encoders

Deep Learning

We have implemented MLP autoencoder, CNN autoencoder, and UNET autoencoder for image to image prediction on Camx dataset.

Hypernet and Neural Processes on CelebA

Deep Learning

Meta Learning

Our goal is constructing the whole image from its few context points.

Active Learning for CNN: A Core-Set Approach

Active Learning

Deep Learning

Here I have implemented Active learning on CIFAR10 dataset using ResNet18 as the base model. We currently use the following active learning strategies: - Diversity…

Towards Scalable Identification of Brick Kilns from Satellite Imagery with Active Learning

Active Learning

Deep Learning

Paper accepted and in nomination for best paper award in NeurIPS 2023 Workshop on ReALML (Active learning and Machine learning in real world)

Active Learning: A Core-Set Approach demo

Active Learning

We want to select the most uncertain samples from the unlabeled pool. However, we also want to ensure that the selected samples are diverse. This is because if we select…