I’m interested in understanding human-level intelligence. Before we can figure out the theory of mind, I believe it’s necessary to understand the patterns of signals that humans are dealing with. So at this moment, I’m working on finding the elementary mathematical structures of natural signals. There are not many of these structures and each of them is not necessarily complex. But, when these simple elements entangle with each other, things quickly exceed the ability of our imagination. My job is to identify and factorize them clearly, concisely and completely, or even uniformly and then mix them together. Thus we can build unsupervised and generative models for the signals, in a somewhat principled way. This might be the first step towards building a truly intelligent machine, which can dream as we do. To be honest, I might be looking for something that doesn’t exist though I suspect there is a hidden gem in the mist, very close.