Abstract. The problem of effciently applying a kernel-induced feature space factorization to a large-scale data sets is addressed in this thesis. Kernel matrix factorization methods have showed good performances solving machine learning and data analysis problems. However, the present growth of the amount of information available implies the problems can not be solved with conventional methods, due their high time and memory requirements. To solve this problem, a new kernel matrix factorization method is proposed called online kernel matrix factorization (OKMF). This method overcomes the time and memory limitations with two strategies. The first is imposing a budget restriction, i.e., restricting the number of samples needed to represent the feature space base. The second is using stochastic gradient descent to compute the factorization, allowing OKMF to scale linearly in time to large-scale data sets. Experimental results show OKMF is competitive with other kernel methods and is capable to scale to a large-scale data sets.