Interactive Machine Learning
Our systems enrich human-machine collaboration to make learning more efficient.
Our scientists are interested in continuous learning through interaction. We develop systems that focus on enriching human-machine collaboration to make learning more efficient. Our interactive learning systems involve humans teaching machines (e.g., human demonstrations), machines teaching machines (e.g., interactive model building), and machines teaching humans (e.g., explaining learned models and their outputs). Current research focuses on improving learning by leveraging domain understanding, incorporating contextual information and human feedback, as well as modeling human perception and reasoning.
Our team’s expertise includes anomaly detection, human-in-the-loop learning, knowledge mining, incorporating physical model intelligence into learning, inverse reinforcement learning, deep learning, and statistical relational learning. We apply these competencies to a diverse range of applications and technologies, including augmented reality assistance, consumer products, IoT analytics, healthcare, railroad fault diagnosis, transportation, and cybersecurity.