Novel machine learning method based on cross-media human common knowledge formation

The invention provides a novel machine learning method based on cross-media human common knowledge formation, and relates to the technical field of machine learning algorithms. The novel machine learning method comprises a deep recording learning module, a central processing module, a quick calling module, an information storage module, an operation module and an information transmission module; the deep recording learning module comprises a logic reasoning unit, an induction learning unit and an intuition pause unit; and the central processing module comprises a logic identification unit, a teaching example identification unit and a distributed processor. A novel model and method for mutual coordination and complementation of symbol logical reasoning, inductive learning and intuitive pause are established through an inductive learning unit, an intuitive pause unit and a logical reasoning unit, inconsistency analysis of cross-media knowledge is supported, learned knowledge is effectively induced through the i