Purva Tendulkar

I am an incoming PhD student at Columbia University, where I will be advised by Prof. Carl Vondrick on topics of computer vision and machine learning. I am interested in developing creative, versatile and personable AI systems.

I completed my Master's in Computer Science at Georgia Tech in 2020 where I was advised by Prof. Devi Parikh. My Master's Thesis is available here. I have also collaborated with Ani Kembhavi at Allen AI. In 2018, I obtained my Bachelor's Degree in Computer Science from College of Engineering Pune (COEP), India.

I've had the pleasure of interning at:

I am trained in Indian classical music and play the harmonium instrument. I also enjoy hiking, photography and sketching!

Email  /  CV  /  Google Scholar  /  Github  /  LinkedIn

SOrT-ing VQA Models : Contrastive Gradient Learning for Improved Consistency
Sameer Dharur, Purva Tendulkar, Dhruv Batra, Devi Parikh, Ramprasaath R. Selvaraju
NAACL, 2021
Interpretable Inductive Biases and Physically Structured Learning, NeurIPS, 2020
arXiv / talk / code / poster / bib

We present a gradient-based interpretability approach to Visual Question Answering (VQA) to determine the questions most strongly correlated with the reasoning question on an image, and use this to evaluate VQA models on their ability to identify the relevant sub-questions needed to answer a reasoning question. Next, we propose a contrastive gradient learning based approach, SOrT, and show improvements in model consistency and visual grounding.

Feel The Music: Automatically Generating A Dance For An Input Song
Purva Tendulkar, Abhishek Das, Aniruddha Kembhavi, Devi Parikh
ICCC, 2020 (Oral)
arXiv / talk / video / dances / code / Tech@Facebook article / press / bib

We present a general computational approach that enables a machine to generate a dance for any input music. We encode intuitive, flexible heuristics for what a 'good' dance is: the structure of the dance should align with the structure of the music. This flexibility allows the agent to discover creative dances. Human studies show that participants find our dances to be more creative and inspiring compared to meaningful baselines. We also evaluate how perception of creativity changes based on different presentations of the dance.

SQuINTing at VQA Models: Interrogating VQA Models with Sub-Questions
Ramprasaath R. Selvaraju, Purva Tendulkar, Devi Parikh, Eric Horvitz, Marco Tulio Ribeiro, Besmira Nushi, Ece Kamar
CVPR, 2020 (Oral)
arXiv / talk / data / bib

We investigate the capabilities of VQA models for solving tasks that differ in nature and in complexity. We notice that existing VQA models have consistency issues -- they answer complex reasoning questions correctly but fail on associated low-level perception sub-questions. We quantify the extent to which this phenomenon occurs by creating a new Reasoning split and collecting Sub-VQA, a new dataset consisting of associated perception sub-questions needed to effectively answer the main reasoning question. Additionally, we propose an approach, SQuINT, which forces models to be right for the right reasons.

Trick or TReAT: Thematic Reinforcement for Artistic Typography
Purva Tendulkar, Kalpesh Krishna, Ramprasaath R. Selvaraju, Devi Parikh
ICCC, 2019 (Oral; Best Presentation Award)
arXiv / talk / demo / code / bib

We present a computational approach for semantic reinforcement called TReAT. Given an input word (e.g. exam) and a theme (e.g. education), the individual letters of the input word are replaced by cliparts relevant to the theme which visually resemble the letters - adding creative context to the potentially boring input word. We use an unsupervised approach to learn a latent space to represent letters and cliparts and compute similarities between the two. Human studies show that participants can reliably recognize the word as well as the theme in our outputs (TReATs) and find them more creative compared to meaningful baselines.

Cloned from here!