Label Learning with Implicit Regularization via Transition Matrix (TMR): A Novel Approach

In this section, we present a novel approach, referred to as the Transition Matrix with Implicit Regularization (TMR), for learning labels, which builds upon previous methods using transition matrices. TMR is an end-to-end model that offers convenience, as it does not rely on assumptions about specific types of label noise. Furthermore, we provide theoretical justification for the proposed approach. We proceed by formulating the method in detail and illustrating it theoretically.