Achievement

IGERT trainee Nicholas Butko has been pursuing computer vision models inspired by human development. Part of this work will be published in the IEEE Journal Transactions on Autonomous Mental Development, and another part will be archived in the Proceedings of the International Conference on Development and Learning. Both of these publications strive to bridge the gap between learning and vision in humans and machines. In one research project, we showed that "information gain" is a powerful source of self-motivation that can be used to learn effective eye-movement strategies, which in turn can be applied to improve the efficiency of state of the art object detection algorithms. In the second, we showed how a robot could use developmental experience to learn the relationship between its sensors (cameras) and actuators (motors) in a way that is independent of its body. This work addresses the computational learning problem faced by human infants in learning how to use their bodies.