Learn to Categorize or Categorize to Learn? Self-Coding for Generalized Category Discovery
This addresses the limitation of traditional supervised models in open-world learning by enabling test-time category discovery, though it appears incremental in its approach to self-supervised methods.
The paper tackles the problem of discovering unknown categories at test time by conceptualizing categories as optimal solutions to optimization problems, proposing a self-supervised method that assigns minimum length category codes to handle fine-grained categories, with experimental results showing efficacy in managing unknown categories and theoretical proof of optimality.
In the quest for unveiling novel categories at test time, we confront the inherent limitations of traditional supervised recognition models that are restricted by a predefined category set. While strides have been made in the realms of self-supervised and open-world learning towards test-time category discovery, a crucial yet often overlooked question persists: what exactly delineates a category? In this paper, we conceptualize a category through the lens of optimization, viewing it as an optimal solution to a well-defined problem. Harnessing this unique conceptualization, we propose a novel, efficient and self-supervised method capable of discovering previously unknown categories at test time. A salient feature of our approach is the assignment of minimum length category codes to individual data instances, which encapsulates the implicit category hierarchy prevalent in real-world datasets. This mechanism affords us enhanced control over category granularity, thereby equipping our model to handle fine-grained categories adeptly. Experimental evaluations, bolstered by state-of-the-art benchmark comparisons, testify to the efficacy of our solution in managing unknown categories at test time. Furthermore, we fortify our proposition with a theoretical foundation, providing proof of its optimality. Our code is available at https://github.com/SarahRastegar/InfoSieve.