Two teams from IIT Hyderabad, under Dr Vineeth N Balasubramanian, Head, Department of Artificial Intelligence, have won the Best Paper Awards at CVPR 21.
The first team, including Puneet Mangla, Vedant Singh and Shreyas Havaldar, won the Best Paper Award at CVPR 21 Workshop on “Adversarial Machine Learning in Real-World Computer Vision Systems and Online Challenges”. The paper was titled ‘On the Benefits of Defining Vicinal Distributions in Latent Space‘.
The second team comprising Abbavaram Gowtham and Benin Godfrey bagged the Best Paper Award at the CVPR 21 Workshop on Causality in Vision. The paper was titled ‘CANDLE: An Image Dataset for CausalANalysis in DisentangLed REpresentations, Workshop on Causality in Vision‘.
Analytics India Magazine got in touch with the IIT Hyderabad teams to get their perspective on winning the prestigious awards.
First team
Puneet Mangla, Vedant Singh and Shreyas Havaldar are Computer Science and Engineering (CSE) students.
The aim of the workshop on “Adversarial Machine Learning in Real-World Computer Vision Systems and Online Challenges (AML-CV)” was to focus on recent research and future directions for security problems in real-world machine learning and computer vision systems. Organisers seek to come to a consensus on a rigorous framework to formulate adversarial machine learning problems in computer vision, characterise the properties that ensure the security of perceptual models, and evaluate the consequences under various adversarial models.
“Well, the news came as a big surprise as it was totally unexpected. Being best among several great submissions is a thought that never occurred to me. So, yeah, the happiness was undeniable. CVPR is a premier conference in the area of Computer Vision, and getting featured in one of its workshops is indeed a cherishable moment. However, it was not the best part. The best part was getting an exclusive opportunity to present our work in the workshop and in front of researchers around the globe. You get to interact with them, discuss your work, get feedback, make connections and get opportunities for collaboration,” Puneet Mangla said.
Meanwhile, Vedant Singh said: “This acceptance came after a couple of rejections, so it feels better than a regular paper acceptance and helps get the confidence back up. Also, the best paper award is the perfect cherry on top.”
“The news of us winning the best paper brought a lot of joy and invigoration. This best paper award motivates me to pursue my ideas passionately, without the fear of failure. It also reassures me that the research I’m pursuing is both challenging and valuable and thus encourages me to devote more efforts to research, exploration and experimentation, especially in the domain of Adversarial ML,” said Shreyas Havaldar.
Second team
The second team, including Abbavaram Gowtham Reddy, 30, PhD, CSE – IITH; and Benin Godfrey L, 21, MSc. PSG College of Technology, Coimbatore, interning at IIT, Hyderabad, presented the paper at CVPR 2021 workshop on “Causality in Vision.”
Causality deals with data generation, model training, and inference with a view to remove the spurious bias, disentangle the desired model effects, and modularise reusable features that generalise well. The workshop aims to provide a comprehensive yet accessible overview of existing causality research and help CV researchers know why and how to apply causality in their own work.
“I am feeling happy, and the best paper award motivates me to continue to produce good quality research work along the direction of applying/using the ideas of Causality in Machine Learning,” said Abbavaram Gowtham Reddy.
“It’s a mix of happiness and motivation. I feel humbled by the response and empowered to continue on with more vigour,” said Benin Godfrey L.
