Learning non-convex abstract concepts with regulated activation networks
- PDF / 2,458,692 Bytes
- 29 Pages / 439.642 x 666.49 pts Page_size
- 61 Downloads / 160 Views
Learning non-convex abstract concepts with regulated activation networks A hybrid and evolving computational modeling approach Rahul Sharma1 · Bernardete Ribeiro1 · Alexandre Miguel Pinto1 · F. Am´ılcar Cardoso1
© Springer Nature Switzerland AG 2020
Abstract Perceivable objects are customarily termed as concepts and their representations (localistdistributed, modality-specific, or experience-dependent) are ingrained in our lives. Despite a considerable amount of computational modeling research focuses on concrete concepts, no comprehensible method for abstract concepts has hitherto been considered. Abstract concepts can be viewed as a blend of concrete concepts. We use this view in our proposed model, Regulated Activation Network (RAN), by learning representations of non-convex abstract concepts without supervision via a hybrid model that has an evolving topology. First, we describe the RAN’s modeling process through a Toy-data problem yielding a performance of 98.5%(ca.) in a classification task. Second, RAN’s model is used to infer psychological and physiological biomarkers from students’ active and inactive states using sleep-detection data. The RAN’s capability of performing classification is shown using five UCI benchmarks, with the best outcome of 96.5% (ca.) for Human Activity recognition data. We empirically demonstrate the proposed model using standard performance measures for classification and establish RAN’s competency with five classifiers. We show that the RAN adeptly performs classification with a small amount of data and simulate cognitive functions like activation propagation and learning. Keywords Computational modeling · Hybrid models · Machine learning · Dynamic models · Abstract concepts · Non-convex models · Evolving topology
Rahul Sharma
[email protected] Bernardete Ribeiro [email protected] Alexandre Miguel Pinto [email protected] F. Am´ılcar Cardoso [email protected] 1
CISUC- University of Coimbra, Coimbra, Portugal
R. Sharma et al.
Mathematics Subject Classification (2010) 91E99
1 Introduction Investigations related to concepts is an essential domain in cognitive and psychological research. In theory, several frameworks represent concepts as process-oriented, symbolic or distributed, and knowledge-based conceptual representations [34]. Usually, these conceptual representations are actualized via mathematical, theoretical (conceptual), or computational cognitive modeling approaches [9]. By definition, a concept refers to an ‘idea’ or a set of ‘ideas’. In a computational domain, a concept can be identified by a feature or a set of features. Concepts are also viewed in the form of a hierarchy where concrete concepts (features) are situated at lower levels and the abstract concepts (features) are at a higher level. Abstracts concepts can be seen as generalization of concrete concepts [48, 59]. There have been efforts to study abstract concepts mathematically [50], and theoretically [12, 13], but are seldom addressed computationally [34]. This article proposes a computational modeling ap
Data Loading...
