Please note: This seminar will take place in DC 1302.
Marta
Kryven,
Postdoctoral
Researcher
Department
of
Brain
and
Cognitive
Sciences
Massachusetts
Institute
of
Technology
Engineering cognitively-inspired AI is crucial for building systems that can understand people and interact with them in safe and productive ways. Successful deployment of such systems in real-world applications (such as autonomous vehicles, service robots and AI assistants) requires accurate computational models of human cognition — and in particular, AI models capable of generating plans and interpreting behavior in human-like ways.
To accomplish this goal, I build computational models of how people plan in realistic scenarios — such as navigating cities and buildings, searching for a lost object, drawing a figure, and forming social attributions based on behaviour. My research connects methods from AI, statistics, neuroscience, and urban planning to validate these models in behavioral experiments, and on real-world data.
Bio: Marta Kryven completed her PhD in Computer Science in 2018 at University of Waterloo, where she built computational models of visual perception. She then continued her research as a postdoc at MIT’s Brain and Cognitive Sciences department, in the Computational Cognitive Science lab advised by Joshua Tenenbaum. Her postdoctoral work focuses on modeling how humans plan, form social attributions based on conjectured models of behavior, and use language when solving problems challenging for modern AI. In spare time she enjoys classical piano and designing educational toys.