Our lab aims to understand the computational basis of intelligence. What makes people smarter than machines? How can we build machines that learn and think in more human-like ways?
There have been remarkable recent advances in machine learning and AI. Computers have beaten Jeopardy champions, defeated Go masters, driven autonomous cars, and shattered records for object and speech recognition. Progress has been impressive, yet today’s AI provides nothing like the general purpose intelligence that we have as humans. Human minds solve a diverse array of computational problems that stump the best machines: learning new concepts, learning new tasks, understanding scenes, compositional learning, asking questions, forming explanations, amongst many others. Machines also struggle to simulate other facets of human intelligence, including creativity, curiosity, self-assessment, and commonsense reasoning.
We study human cognitive abilities that elude the smartest machines. Almost by definition, these domains are important pursuits for both cognitive science and data science. In cognitive science, if people have abilities that existing algorithms do not, we must understand their computational basis. In data science and machine learning, these distinctively human abilities are both important open problems as well as opportunities to reverse-engineer the human solutions.
Our work demands a combination of cognitive modeling, behavioral experiments, and machine learning research. Through this interdisciplinary approach, our work has revealed key cognitive ingredients that people use but are missing in contemporary machine learning. It has also led to new data science and machine learning techniques inspired by the cognitive solutions to difficult computational problems.
See the Projects page for examples of our current research directions.