Artificial intelligence (AI) is a strong area of cross-disciplinary research spanning mathematics, robotics, sensory systems, computation, and neuroscience. We are working on the foundations of the field and we are applying AI methods to the sciences and to planetary exploration.
Machine learning is a foundational discipline for AI. Anima Anandkumar works on deep learning methods that are generalizable and robust, including those for solving partial differential equations and other scientific simulations, reinforcement learning for control, and tensor methods. Adam Wierman focuses on online learning and online optimization, with applications to networked systems.
Safety is a critical aspect of AI. The intersection of learning and control, and the application of such policies to safety-critical systems such as autonomous vehicles and power systems (Anima Anandkumar, Babak Hassibi, Eric Mazumdar, Adam Wierman, and Yisong Yue).
When AI systems interact with people it is crucial that they are safe and fair. Defining fairness criteria and measuring bias in artificial systems is fundamental. Pietro Perona develops experimental methods for measuring bias in AI systems that may be used in critical Agosto4! applications.
Vision is a rich sensory system that helps humans animals and machines in understanding the world, navigation, manipulation and social interaction. Katie Bouman, Pietro Perona and Yisong Yue study the computational foundations of vision, and the application of vision to the analysis of images in science, medicine and industry.
Biology developed intelligent autonomous systems spanning a large range of power / performance trade-offs. Pietro Perona and Yisong Yue collaborate with neuroscientists to understand the structure of animal behavior, the role and computational value of emotion in behavior and how animals learn new tasks. Pietro Perona collaborates with Caltech psychologists and social scientists to understand the nature of search and decision in humans.
Caltech is at the forefront of applying AI to the sciences. The AI4Science initiative was founded by Yisong Yue and Anima Anandkumar with the mission of developing AI for scientific domains, and it has seen many success stories of interdisciplinary collaborations across campus.