Unified and scalable learning in multimedia information retrieval

Statistical-learning approaches such as unsupervised learning, supervised learning, active learning, and reinforcement learning have generally been separately studied and applied to solve application problems. In this talk, I will present our recent work on a unified learning paradigm (ULP). ULP is motivated by how human being acquires knowledge: we learn by being taught (supervised learning), by self-study (unsupervised learning), by asking questions (active learning), and by being examined for the ability to generalize (reinforcement learning). I will present our recent ICML and KDD work on ULP, which can substantially reduce the amount of required training data. I will also present our proposed algorithmic and data processing techniques for speeding up kernel-based learning (ICML, KDD, SIAM, and MM) for multimedia information retrieval. Finally, I will touch basis on my work at Google that relates to the multimedia community.