How to Build a Feeling Machine: Homeostasis, Multi-sensory Integration, and Soft Robotics
Date: 2020/12/10 - 2020/12/10
Academic Seminar: How to Build a Feeling Machine: Homeostasis, Multi-sensory Integration, and Soft Robotics
Speaker: Dr. Kingson Man
Time: 9:00 a.m.-10:00 a.m., December 10th, 2020 (Beijing Time)
Location: via Zoom (Meeting ID: 64060472084, Password: 7680)
Abstract
The dream of creating artifacts with a capacity for human-like feeling has been with us since antiquity. Although some existing AI systems can display extreme performance on narrow tasks, none possess the broad intelligence and behavioral repertoire of living creatures. Creative behavior is motivated and evaluated by feelings of the state of life inside the body. This talk presents work on how an analogue of feelings can be constructed in machines that are vulnerable to the environment and that consequently follow the dictates of homeostasis. In cross-disciplinary work, I integrate recent breakthroughs from i) soft robotics and ii) deep learning for multi-sensory integration. The young field of soft robotics presents exciting new opportunities for machines to generate homeostatic data that are far richer than those achievable in rigid-bodied robots. The tools of deep learning are applied to build brain-inspired multi-sensory architectures, bridging across internal and external worlds to create a sense of “self”. I present results from human brain-imaging studies and neural network simulations, concluding with a vision for artificial affective intelligence. This work has the potential to domesticate the brute industrial robots of today, turning them into our cherished companions that possess an analogue of feelings and, perhaps, empathy.
Biography
Dr. Kingson Man earned his PhD in Neuroscience from the University of Southern California, mentored by Antonio Damasio. He was previously a research intern at NASA Jet Propulsion Laboratory, in the Mission Control Systems - Deep Learning Technologies Group. He is currently a Senior Research Associate at the USC Brain and Creativity Institute, where he leads research into feelings and intelligence, both natural and artificial.