Speaker: Dr Yang Haiqin
Chinese University of Hong Kong
Dr Yang presented his research results in artificial intelligence. Nowadays, information overload becomes a critical issue for people’s daily life. To tackle this problem, recommender system is proposed and is promising to provide favorite personalised recommendation or services. Collaborative filtering (CF) techniques, making prediction of users’ preference based on users’ previous behaviors, have become one of the most successful technologies to build modern recommender systems. Several challenging issues occur in previously proposed CF methods: 1) most CF methods ignore users’ response patterns and may yield biased parameter estimation and suboptimal performance;2) some CF methods adopt heuristic weight settings, which lacks a systematic implementation; 3) the multinomial mixture models may weaken the computational ability of matrix factorization for generating the data matrix, thus increasingthe computational cost of training. To resolve these issues, Dr Yang incorporated users’ response models into the probabilistic matrix factorization (PMF), a popular matrix factorisation CF model, to establish the Response Aware Probabilistic Matrix Factorization(RAPMF) framework. More specifically, he made the assumption on the user response as a Bernoulli distribution which is parameterized by the rating scores for the observed ratings while as a step function for the unobserved ratings. Moreover, Dr. Yang sped up the algorithm by a mini-batch implementation and a crafting scheduling policy. Finally, he conducted thorough empirical evaluation on both synthetic and real-world data sets to demonstrate the merits of the proposed RAPMF and its mini-batch implementation.