Meta-Unsupervised-Learning: A supervised approach to unsupervised learning
This provides a new foundational approach to unsupervised learning, potentially impacting all of ML/AI by addressing core limitations.
The paper tackles the subjectivity in unsupervised learning by reducing it to supervised learning using prior knowledge from heterogeneous tasks, demonstrating versatility across clustering, outlier detection, and similarity prediction with theoretical bounds that circumvent Kleinberg's impossibility theorem.
We introduce a new paradigm to investigate unsupervised learning, reducing unsupervised learning to supervised learning. Specifically, we mitigate the subjectivity in unsupervised decision-making by leveraging knowledge acquired from prior, possibly heterogeneous, supervised learning tasks. We demonstrate the versatility of our framework via comprehensive expositions and detailed experiments on several unsupervised problems such as (a) clustering, (b) outlier detection, and (c) similarity prediction under a common umbrella of meta-unsupervised-learning. We also provide rigorous PAC-agnostic bounds to establish the theoretical foundations of our framework, and show that our framing of meta-clustering circumvents Kleinberg's impossibility theorem for clustering.