Landskape

a tiny research group on the edge of theoretical deep learning

Find out more

Projects

AP/GP: Anytime Progressive Growing + Pruning

Progressive Growing and Pruning for ALMA.

PSNR: Progressive Scale Noisy Reconstruction

Progressive scale noisy reconstruction for image denoising using ALMA framework.

Reprogrammers are robust learners

Proving reprogramming preserves robustness upon transfer.

Publications

APP: Anytime Progressive Pruning

Diganta Misra , Bharat Runwal , Tianlong Chen , Zhangyang Wang , Irina Rish

Preprint

April, 2022

Details PDF Code Cite WandB dashboard Neural Scaling Laws course presentation MLC Research Jam #8

Rotate to Attend: Convolutional Triplet Attention Module

Diganta Misra , Trikay Nalamada , Ajay Uppili Arasanipalai , Qibin Hou

Winter Conference on Applications of Computer Vision (WACV)

January, 2021

Details PDF Slides Video Code Cite

Mish: A Self Regularized Non-Monotonic Activation Function

Diganta Misra

31st British Machine Vision Conference (BMVC)

September, 2020

Details PDF Video Code Cite SOTA in Object Detection on COCO-testdev BMVC 2020 Machine Learning Cafe Podcast Sicara Deep Club Computer Vision Talks WandB dashboard

Team

Diganta Misra

Founder and Director, Landskape AI

Himanshu Arora

Director, Landskape AI

Trikay Nalamada

Director, Landskape AI

Ajay Uppili Arasanipalai

Director, Landskape AI

Bharat Runwal

Student Researcher, Landskape AI

Richeek Das

Student Researcher, Landskape AI

Javier Ideami

Advisor, Landskape AI

Federico Andres Lois

Advisor, Landskape AI

Contact

In collaboration with:

Continual AI

Weights & Biases

Supported by contributions from:

NCSA

OpenPOWER Foundation