This paper introduces a new metamodel-based knowledge representation that\nsignificantly improves autonomous learning and adaptation. While interest in\nhybrid machine learning / symbolic AI systems leveraging, for example,\nreasoning and knowledge graphs, is gaining popularity, we find there remains a\nneed for both a clear definition of knowledge and a metamodel to guide the\ncreation and manipulation of knowledge. Some of the benefits of the metamodel\nwe introduce in this paper include a solution to the symbol grounding problem,\ncumulative learning, and federated learning. We have applied the metamodel to\nproblems ranging from time series analysis, computer vision, and natural\nlanguage understanding and have found that the metamodel enables a wide variety\nof learning mechanisms ranging from machine learning, to graph network analysis\nand learning by reasoning engines to interoperate in a highly synergistic way.\nOur metamodel-based projects have consistently exhibited unprecedented\naccuracy, performance, and ability to generalize. This paper is inspired by the\nstate-of-the-art approaches to AGI, recent AGI-aspiring work, the granular\ncomputing community, as well as Alfred Korzybski's general semantics. One\nsurprising consequence of the metamodel is that it not only enables a new level\nof autonomous learning and optimal functioning for machine intelligences, but\nmay also shed light on a path to better understanding how to improve human\ncognition.\n