, Unsupervised Both Strategy 4 and Strategy 3 squeeze the X-form from inside to high abstraction. However, Strategy 4 uses unsupervised data (at least partially), this is very desirable. But, using unsupervised data, we need to have a stronger capability. Just like strategy 2, strategy 4 can achieve universal learning machine as well. Now we know universal learning machine is achievable, Strategy 4 -Squeeze X-form to Higher Abstractions

C. Xiong, Discussion on Mechanical Learning and Learning Machine, 2016.

C. Xiong, Descriptions of Objectives and Processes of Mechanical Learning

C. Xiong, What Really is Deep Learning Doing

C. Xiong, Chinese patent application Application #

P. Domingos, The Master Algorithm, Talks at Google

L. Valiant, A theory of the learnable, Communications of the ACM, vol.27, 1984.

P. Wang, A Logical Model of Intelligence An introduction to NARS

. Zhong-yixin, Mechanism-based artificial intelligence theory, CAAI transactions on intelligent systems, vol.13, p.218, 2018.

. He-huacan, Universal logic theory: logical foundation of mechanism-based artificial intelligence theory, CAAI transactions on intelligent systems, vol.13, p.1936, 2018.

. Wang-peizhuang, Factor space-mathematical basis of mechanism based artificial intelligence theory, CAAI transactions on intelligent systems, vol.13, p.3754, 2018.