We study the problem of inverse reinforcement learning (IRL) with the added twist that the learner is assisted by a helpful teacher. More formally, we tackle the following question: How could a teacher provide an informative sequence of demonstrations to an IRL agent to speed up the learning process? We prove rigorous convergence guarantees of a new iterative teaching algorithm that adaptively chooses demonstrations based on the learner’s current performance. Extensive experiments with a car driving simulator environment show that the learning progress can be speeded up drastically as compared to an uninformative teacher.MPI, Saarbrucken, Germany
The ability of intelligent agents to play games in human-like fashion is popularly considered a benchmark of progress in Artificial Intelligence. In our work, we introduce the first computational model aimed at Pictionary, the popular word-guessing social game. We first introduce Sketch-QA, a guessing task. We analyze the resulting dataset and present many interesting findings therein. To mimic Pictionary-style guessing, we propose a deep neural model which generates guess-words in response to temporally evolving human-drawn object sketches. Our model even makes human-like mistakes while guessing, thus amplifying the human mimicry factor. We evaluate our model on the large-scale guess-word dataset generated via Sketch-QA task and compare with various baselines.IISc Bangalore, India
This paper addresses detecting, localizing and recognizing various grocery products in retail store images. Our object recognition algorithm achieves this goal using just one image per product for training, assuming that the category of the products (like cereals, rice, etc.) is known. This algorithm uses logo detection as a precursor to product recognition. So, the first step involves detecting and classifying products, at a broader level, based on their brands. The second step is the finer classification step for recognizing and localizing the exact product label, which involves using colour information. This hierarchical approach limits the confusion in classifying similar looking products and outperforms product recognition that was implemented without logo detection.IIIT Bangalore, India
Existing DBMS's are equipped with syntactical checkers. But, often the queries do not produce the expected results because of some logical errors. These errors can often go unnoticed. We developing a prototype for a visual debugger to ease the debugging of recursive and long SQL queries. Our debugger uses informative visualisations like Query Node Graphs. We also incorporate provenance, another important aspect which can aid the degugging process.University of Maryland, College Park
VideoKen is an online platform for educational videos. It enables the learning process by adding features that will help learners while watching educational videos. Our work involved understanding what led to disengagement of users while watching the educational videos. Also, what were possible interventions the system could provide to engage the users back.IIIT Bangalore, India
Developed a prototype on Intel Edison platform for a smart shoe which establishes the relationship between the physical inactivity and exposure to polluted environment. The project involved both software and hardware experise. Incorporated accelerometers, pressure sensors and gas sensors and used naive machine learning algorithms like k-Nearest Neighbours and Decision Trees.IIIT Bangalore, India
Interned under Prof. Adish Singla and worked on problems at the intersection of Inverse Reinforcement Learning and Machine Teaching.
Guided by Prof. Venkatesh Babu and worked with Ravikiran Sarvadevbatla and Shiv Surya to develop computational models for Pictionary-style word guessing.