Real-Time Collaborative Filtering Using Extreme Learning Machine
Wanyu Deng, Qinghua Zheng, Lin Chen · 2009
Because of long-consuming training or similarity computing, most traditional collaborative filtering algorithms are off-line methods and can’t be applied in collaborative-filtering services that have accumulated large amounts of data and need to compute predictions under real-time conditions. In order to address this problem, we propose a novel real-time collaborative filtering algorithm, called RCF, based on Extreme Learning Machine (ELM). The initial training and updating of RCF are very fast and can be finished in real time. The experimental results show that the mean recommendation time of RCF is shorter than SVD/ANN and correlation-based algorithms reported in other papers while the accuracy is better.