Implicit Regularization in Matrix Factorization

This generalization ability cannot be explained by the capacity of the explicitly specified model class (namely, the functions representable in the chosen architecture). Instead, it seems that the optimization algorithm biases us toward a "simple" model, minimizing