Litu Rout

I am a second year PhD student at The University of Texas Austin, advised by Prof. Constantine Caramanis and Prof. Sanjay Shakkottai. My current research focuses on the theoretical foundations of diffusion models, generative modeling, posterior sampling, and stochastic optimization. I am currently working as a student researcher at Google Research.

Prior to UT Austin, I worked as a Scientist/Engineer-SD at the Indian Space Research Organisation, where I developed operational deep learning algorithms and analyzed their convergence properties.

I received my BTech from the Indian Institute of Space Science and Technology (IIST). I was fortunate to be advised by Prof. Rama Krishna Gorthi and Prof. Deepak Mishra during my undergraduate research.

Contact  /  Google Scholar  /  DBLP  /  LinkedIn  /  GitHub

profile photo
Updates

Reviewer

Neurips (2022, 2023, 2024), ICLR (2022,2023,2024), ICML (2022,2023,2024), CVPR (2024), TMLR (2024), AISTATS (2023), Pattern Recognition (2022), NeurIPS ML4PS (2021).

Publications
Diffusion Probabilistic Models
NEW   RB-Modulation: Training-Free Personalization of Diffusion Models using Stochastic Optimal Control
Litu Rout, Yujia Chen, Nataniel Ruiz, Abhishek Kumar, Constantine Caramanis, Sanjay Shakkottai, and Wen-Sheng Chu
Preprint, 2024
Project Page   PDF   ArXiv     Code

We introduce Reference-Based Modulation (RB-Modulation), a training-free plug-and-play solution for content and style personalization. By incorporating style features into the controller’s terminal cost, we modulate the drift field in diffusion models’ reverse dynamics, enabling training-free personalization. Further, we propose an Attention Feature Aggregation (AFA) module that decouples content from the reference style image.

NEW   Beyond First-order Tweedie: Solving Inverse Problems using Latent Diffusion
Litu Rout, Yujia Chen, Abhishek Kumar, Constantine Caramanis, Sanjay Shakkottai, and Wen-Sheng Chu
CVPR, 2024
Project Page   PDF   ArXiv   Talk   Code

We present an efficient second-order approximation using Tweedie's formula to mitigate the bias incurred in the widely used first-order samplers. With this method, we devise a surrogate loss function to refine the reverse process at every diffusion step to address inverse problems and perform high-fidelity text-guided image editing.

Solving Linear Inverse Problems Provably via Posterior Sampling with Latent Diffusion Models
Litu Rout, Negin Raoof, Giannis Daras, Constantine Caramanis, Alex Dimakis, and Sanjay Shakkottai
NeurIPS, 2023
OpenReview   Poster   ArXiv   Code   Demo  Presentation

Solving inverse problems (e.g. inpainting/deblurring) for general domain images is hard. Magic Eraser and other commercial tools use separately trained models for each task. We introduce PSLD, a method that uses Stable Diffusion to solve all linear problems without any extra training.

A Theoretical Justification for Image Inpainting using Denoising Diffusion Probabilistic Models
Litu Rout, Advait Parulekar, Constantine Caramanis, and Sanjay Shakkottai
UT Austin Technical Report, 2023
ArXiv

We provide a theoretical justification for sample recovery using diffusion based image inpainting in a linear model setting. Unlike most inpainting algorithms, we prove that diffusion based inpainting generalizes well to unseen masks without retraining. Motivated by our analysis, we propose a modified RePaint algorithm we call RePaint+ that provably recovers the underlying true sample and enjoys a linear rate of convergence.

Stochastic Optimization
Beyond Uniform Smoothness: A Stopped Analysis of Adaptive SGD
Matthew Faw*, Litu Rout*, Constantine Caramanis, and Sanjay Shakkottai
COLT, 2023   (* Equal contribution)
ArXiv

We develop a technique that allows us to prove convergence rates for (L0, L1)-smooth functions without assuming uniform bounds on the noise support. The key innovation behind our results is a carefully constructed stopping time. This is simultaneously large on average and allows us to decorrelate the adaptive stepsizes from the gradients, which is a major challenge in many analyses.

Optimal Transport
Generative Modeling with Optimal Transport Maps
Litu Rout, Alexander Korotin, and Evgeny Burnaev
ICLR, 2022
OpenReview   PDF   ArXiv   Slides   Poster   Code

While Optimal Transport (OT) cost serves as the loss for popular generative models, we demonstrate that the OT map can be used as the generative model itself. Previous analogous approaches consider OT maps as generative models only in the latent spaces due to their poor performance in the original high-dimensional ambient space. In contrast, we fit OT maps directly in the ambient space, e.g., a space of high-dimensional images.

Hierarchical Sliced Wasserstein Distance
Khai Nguyen, Tongzheng Ren, Huy Nguyen, Litu Rout, Tan Nguyen, and Nhat Ho
ICLR, 2023
OpenReview   PDF   ArXiv

A major concern of Sliced Wasserstein (SW) distance is that it requires a large number of projections in high-dimensional settings. To address this concern, we derive projections from a small number of bottleneck projections. We introduce Hierachical Radon Transform (HRT) that recursively applies Radon Transform (RT). We design Hierarchical Sliced Wasserstein (HSW) distance to estimate the discrepancy between measures in high dimensions.

Unpaired Image Super-Resolution with Optimal Transport Maps
Milena Gazdieva*, Litu Rout*, Alexander Korotin*, Alexander Filippov, and Evgeny Burnaev
Preprint, 2022   (* Equal contribution)
ArXiv

First, we prove that GANs with content or identity losses learn optimal transport (OT) maps between source and target measures in super-resolution tasks. Second, we empirically demonstrate that these learned OT maps are biased and provide an OT solver to recover an unbiased OT map. It provides nearly state-of-the-art performance on the unpaired AIM19 benchmark without having to use content or identity losses.

Neural Network Learning Theory & Pattern Recognition
Why Adversarial Interaction Creates Non-Homogeneous Patterns: A Pseudo-Reaction-Diffusion Model for Turing Instability
Litu Rout
AAAI (acceptance rate: 21%), 2021 (Extended Version)
PDF   ArXiv   Slides   Poster   Teaser   Presentation

In this paper, we intend to demystify an interesting phenomenon: adversarial interaction in GANs creates non-homogeneous equilibrium by inducing Turing instability in a Pseudo-Reaction-Diffusion (PRD) model. This is in contrast to supervised learning where the identical model achieves homogeneous equilibrium.

Towards A Pseudo-Reaction-Diffusion Model for Turing Instability in Adversarial Learning
Litu Rout
NeurIPS ML4PS Workshop, 2020 (Short Version)
PDF   Poster   ArXiv

In this study, we observe that a system in which a generator and a discriminator adversarially interact with each other exhibits Turing-like patterns in the hidden layer and top layer of the generator.

Understanding the Role of Adversarial Regularization in Supervised Learning
Litu Rout
CVPR Adversarial ML Workshop, 2021
PDF

Despite numerous attempts sought to provide empirical evidence of adversarial regularization outperforming sole supervision, the theoretical understanding of such phenomena remains elusive. In this study, we aim to resolve whether adversarial regularization indeed performs better than sole supervision at a fundamental level.

Generative Adversarial Networks
ALERT: Adversarial Learning with Expert Regularization using Tikhonov Operator for Missing Band Reconstruction
Litu Rout
TGRS (impact factor: 5.85), 2020
PDF   Preprint

In this article, we devise a method, which we call ALERT, to tackle missing band reconstruction. The proposed method reconstructs missing band with the sole supervision of spectral and spatial priors.

S2A: Wasserstein GAN with Spatio-Spectral Laplacian Attention for Multi-Spectral Band Synthesis
Litu Rout, Indranil Misra, S Manthira Moorthi, and Debajyoti Dhar
CVPR Earth Vision Workshop, Oral Talk (acceptance rate: 26%), 2020
PDF   ArXiv   Slides  Presentation

This paper seeks to address synthesis of high resolution multi-spectral satellite imagery using adversarial learning. Guided by the discovery of attention mechanism, we regulate the process of band synthesis through spatio-spectral Laplacian attention.

Satellite Image Processing
Monte-Carlo Siamese Policy on Actor for Satellite Image Super Resolution
Litu Rout, Saumyaa Shah, S Manthira Moorthi, and Debajyoti Dhar
CVPR Earth Vision Workshop, Oral Talk (acceptance rate: 26%), 2020
PDF   ArXiv   Slides   Presentation

In this study, we propose to parameterize action variables by matrices, and train our policy network using Monte-Carlo sampling. We study the implications of parametric action space in a model-free environment from theoretical and empirical perspective.

Phobos Image Enhancement using Unpaired Multi-Frame Acquisitions from Indian Mars Color Camera
Indranil Misra, Litu Rout, Sunita Arya, Yatharath Bhateja, S Manthira Moorthi, and Debajyoti Dhar
Planetary and Space Science , 2021
PDF   Preprint

This paper describes the techniques developed to enhance the Phobos image from MCC multi-frame acquisitions using image rectification and topographic data. After incorporating these techniques, the final Phobos image appears more representative, spatially enhanced, and has normalized radiometry to study its surface features.

Visual Object Tracking
The tenth visual object tracking vot2018 challenge results
Matej Kristan, Ales Leonardis, Jiri Matas, Michael Felsberg, Roman, Pflugfelder, Litu Rout and others
ECCV VOT Workshop, 2022
PDF

Results of over ninety trackers are presented; many are state-of-the-art trackers published at major computer vision conferences or in journals in the recent years.

The seventh visual object tracking vot2018 challenge results
Matej Kristan, Ales Leonardis, Jiri Matas, Michael Felsberg, Roman, Pflugfelder, Litu Rout and others
ICCV VOT Workshop, 2019
PDF

Results of 81 trackers are presented; many are state-of-the-art trackers published at major computer vision conferences or in journals in the recent years.

Learning Rotation Adaptive Correlation Filters in Robust Visual Object Tracking
Litu Rout, Priya Mariam Raju, Deepak Mishra, and Rama Krishna Gorthi
ACCV, 2018
PDF   ArXiv   Slides

Here, we propose a robust framework that offers the provision to incorporate illumination and rotation invariance in the standard Discriminative Correlation Filter (DCF) formulation. We also supervise the detection stage of DCF trackers by eliminating false positives in the convolution response map.

WAEF: Weighted Aggregation with Enhancement Filter for Visual Object Tracking
Litu Rout, Deepak Mishra and Rama Krishna Gorthi
ECCV VOT Workshop, 2018
PDF

This paper discusses a novel approach to regress in temporal domain, based on weighted aggregation of distinctive visual features and feature prioritization with entropy estimation in a recursive fashion.

The sixth visual object tracking vot2018 challenge results
Matej Kristan, Ales Leonardis, Jiri Matas, Michael Felsberg, Roman, Pflugfelder, Litu Rout and others
ECCV VOT Workshop, 2018
PDF

Results of over eighty trackers are presented; many are state-of-the-art trackers published at major computer vision conferences or in journals in the recent years.

Rotation Adaptive Visual Object Tracking with Motion Consistency
Litu Rout, Sidhartha, Deepak Mishra, and Rama Krishna Gorthi
WACV, 2018
PDF   ArXiv   Code   Slides   Poster  Presentation

In this paper, we study the necessity to capture various physical constraints through motion consistency which has been demonstrated to improve accuracy, robustness and more importantly rotation adaptiveness.

Application of image enhancement and mixture of Gaussian approach in combustion research
Litu Rout, Rajesh Sadanandan and Deepak Mishra
Sadhana, Indian Academy of Sciences, 2019
PDF   IAS

The developed algorithm has been implemented to yield the physically significant chemiluminescence emission from hydroxyl radicals in flames from line-of-sight integrated images. The effectiveness of this algorithm is highlighted using exemplary OH chemiluminescence images captured from a standard swirl stabilized research burner.

Patent
A Method for Sequential Information Condensation using Fourier Basis
Tapan Misra, Litu Rout
SAC, ISRO, 2020
Patent No. 346206, Application No. 202041004166

The present embodiment proposes an efficient Fast Fourier Transform (FFT) based hyper-spectral image compression technique to store multiple acquisitions over same region of interest and thereby, improve Signal to Noise Ratio (SNR) of hyper-spectral images which usually have coarse spatial resolution.

Chapters
Learning Rotation Adaptive Correlation Filters in Robust Visual Object Tracking
Litu Rout, Priya Mariam Raju, Deepak Mishra, and Rama Krishna Gorthi
Computer Vision – ACCV, 2018

In this chapter, we propose a robust framework that offers the provision to incorporate illumination and rotation invariance in the standard Discriminative Correlation Filter (DCF) formulation. We also supervise the detection stage of DCF trackers by eliminating false positives in the convolution response map.

WAEF: Weighted Aggregation with Enhancement Filter for Visual Object Tracking
Litu Rout, Deepak Mishra and Rama Krishna Gorthi
Computer Vision – ECCV Workshops, 2018

This chapter discusses a novel approach to regress in the temporal domain, based on weighted aggregation of distinctive visual features and feature prioritization with entropy estimation in a recursive fashion.

Website Template