Unsupervised Cognition

Abstract: Unsupervised learning methods have a soft inspiration in cognition models. To this day, the most successful unsupervised learning methods revolve around clustering samples in a mathematical space. In this paper we propose a primitive-based, unsupervised learning approach for decision-making inspired by a novel cognition framework. This representation-centric approach models the input space constructively as a distributed hierarchical structure in an input-agnostic way. We compared our approach with both current state-of-the-art unsupervised learning classification, with current state-of-the-art small and incomplete datasets classification, and with current state-of-the-art cancer type classification. We show how our proposal outperforms previous state-of-the-art. We also evaluate some cognition-like properties of our proposal where it not only outperforms the compared algorithms (even supervised learning ones), but it also shows a different, more cognition-like, behaviour.

• The paper introduces a primitive-based unsupervised learning approach that constructs hierarchical representations from simple primitives.
• The paper applies a cognition-inspired decision-making framework that outperforms traditional unsupervised and some supervised methods in benchmark tasks.
• The paper demonstrates cognition-like properties that enable flexible, robust information processing akin to human learning.

"Unsupervised Cognition" explores unsupervised learning techniques inspired by cognitive models, which seek to mirror some aspects of human learning without direct supervision. The paper proposes a novel approach that constructs a distributed hierarchical structure to represent input data in a way that is agnostic to the specific nature of the inputs. This allows the system to organize and process information more flexibly, similar to certain cognitive processes.

Key contributions of the paper include:

1. Primitive-Based Unsupervised Learning: The method centers around a representation-centric approach that builds complex structures from simpler primitives. This contrasts with traditional methods that often focus on clustering in mathematical space alone.
2. Decision-Making Framework: The approach is applied to decision-making tasks, demonstrating potential advantages over classical techniques by leveraging cognition-inspired models.
3. Performance Evaluation: The authors benchmark their method against state-of-the-art unsupervised learning techniques and supervised methods in specific contexts, such as cancer type classification. Their findings suggest that the proposed method not only performs better than existing unsupervised approaches but also surpasses some supervised algorithms.
4. Cognition-Like Properties: Beyond performance metrics, the proposed method exhibits behavior that is described as more cognition-like. This suggests potential for developing systems that can mimic certain human-like ways of processing and understanding information.

Overall, the paper makes a compelling case for advancing unsupervised learning through models that take inspiration from cognitive science, potentially leading to more robust and versatile AI systems.