Continual Learning of Visual Concepts for Robots through Limited Supervision

For many real-world robotics applications, robots need to continually adapt\nand learn new concepts. Further, robots need to learn through limited data\nbecause of scarcity of labeled data in the real-world environments. To this\nend, my research focuses on developing robots that continually learn in dynamic\nunseen environments/scenarios, learn from limited human supervision, remember\npreviously learned knowledge and use that knowledge to learn new concepts. I\ndevelop machine learning models that not only produce State-of-the-results on\nbenchmark datasets but also allow robots to learn new objects and scenes in\nunconstrained environments which lead to a variety of novel robotics\napplications.\n