Abstract:
When do we have to exploit causal knowledge, and when does associational information suffice? Can we find the causal direction between two variables by analyzing their observed values? Can we figure out where latent variables should be and how they are related? For the purpose of understanding and manipulating systems properly, people often attempt to answer such causal questions. In the past decades, interesting advances were made in fields including machine learning, statistics, and philosophy in order to find causal relations by data analysis. Furthermore, we are also often concerned with artificial intelligence (AI) questions in complex environments. For instance, how can we do transfer learning in a principled way? How can machines avoid adversarial attacks? Interestingly, it has recently been shown that causal information can facilitate understanding and solving various AI problems. This talk focused on how to learn causal representations from observation data and why and how the causal perspective allows adaptive prediction and a potentially higher level of artificial intelligence.
Bio:
Kun Zhang is an associate professor of philosophy and an affiliate faculty in the machine learning department of Carnegie Mellon University, and a senior research scientist at Max Planck Institute for Intelligent Systems, Germany. He is interested in the connection between causality and machine intelligence, and has been actively developing methods for automated causal discovery from various kinds of data and investigating machine learning problems, including transfer learning, adversarial vulnerability, and deep learning, from a causal view. His work has been widely published in major artificial intelligence and machine learning venues. Dr. Zhang coauthored a best student paper for UAI and a best finalist paper for CVPR, and received the best benchmark award of the causality challenge, and has been frequently serving as a senior area chair, area chair, or senior program committee member for most major conferences in machine learning or artificial intelligence, including NeurIPS, ICML, UAI, IJCAI, ICLR, and AISTATS.