CSIS Seminar

Human versus Machine Perception of Patterns - A visual Turing Test: “Are you a human or a robot?”

Abstract

Regularities with varying form and scale pervade our natural and man-made world. From insects to mammals, the ability to sense regular patterns has a neurobiological basis and has been observed in many levels of intelligence and behavior. From Felix Kleinʼs Erlanger program, D’Arcy Thompson’s Growth-and-Form, to the Gestalt principles of perception, much of our understanding of the world is based on the perception and recognition of repetitions, generalized by the mathematical concept of symmetry and symmetry groups. Given the ubiquity of symmetry in both the physical and the digital worlds, a computational model for symmetry-based regularity perception is especially pertinent to computer vision, computer graphics, robotics and machine intelligence in general, where an intelligent being (e.g. a robot) seeks to perceive, reason and interact with the chaotic world in the most effective and efficient manner. Surprisingly, we have limited knowledge on how humans perceive regular patterns and little progress has been made in computational models for noisy, albeit near-regular patterns in real data. In this talk, I present parallels as well as differences between machine perception and human perception of visual regularity. I shall report our recent exploration on human perception of patterns using 2D crystallographic groups as a guiding principle via neuroimaging, an un-supervised learning approach for recurring patterns discovery in the wild, a deep learning framework to mimic human perception of symmetry from photos as well as a successful attempt at building a symmetry-based Turing test to tell humans and robots apart: a symmetry reCAPTCHA.

Speaker Bio

Dr. Yanxi Liu is a Professor of EECS at Penn State University (PSU), University Park, USA, trained in physics/EE, computer science and theoretical robotics/AI (B.S, China; Ph.D. USA; Postdoc, France). With an NSF (USA) research-education fellowship award, she spent one year at DIMACS (NSF center for DIscrete MAthematics and Theoretical Computer Science) before joining the faculty of the Robotics Institute, Carnegie Mellon University for ten years. Currently at PSU, she is the director of the Human Motion Capture Lab for Smart Health and co-directs the Lab for Perception, Action and Cognition (LPAC). Dr. Liu has been a visiting professor at Stanford University, ETH Zurich, Tsinghua University and Google/MSR/MSRA. She is on sabbatical 2023-2024 at CMU. A central theme of Dr. Liu's research is on group theory-based “computational regularity” for multimodality data (funded continuously by US NSF, including a prestigious multidisciplinary INSPIRE grant) with diverse applications in robotics, human/machine perception, human activity in sports and in health. Dr. Liu chaired three international competitions at CVPR, ECCV, ICCV on Computer Vision algorithms for Detecting Symmetry in the Wild, and is the lead author for the book on “Computational Symmetry in Computer Vision and Computer Graphics”. Her industrial visits to Google Mountain View and Microsoft Silicon Valley/Uber resulted in two granted US patents. She has served as a program chair for Computer Vision and Pattern Recognition (CVPR) Conference 2017 and Winter Conference on Applications of Computer Vision (WACV) 2019, area chairs for all major computer vision/graphics conferences (CVPR/ECCV/ICCV/MICCAI/ACM MM/SIGGRAPH), and as an associate editor for IEEE Transaction of Pattern Analysis and Machine Intelligence (PAMI).