In real-world systems, organisations, their behaviours, data and outcomes are embedded with complicated interactions, coupling relationships, and heterogeneities, forming non-IIDness (Independent and Identically Distributed) which goes beyond the classic IID assumptions. This generally applies to any physical, social, economic, virtual, human-made, and hidden systems and organisations. Of the many system complexities, such non-IIDnesses are arguably more fundamental, complex and challenging to understand, quantify and compute, ranging from system constituents, subsystems, environments to the whole-of-system.
In this talk, Prof Cao Longbing will briefly share his preliminary thoughts on the underlying problems and challenges in non-IIDness, limitations of the existing quantitative and computing paradigms, and some outlooks for foundational developments that go beyond existing general thinking, methodologies, and frameworks in statistics, artificial intelligence (AI), data science, and shallow to deep machine learning, etc.