Exploring Geometric Feature Hyper-Space in Data to Learn Representations of Abstract Concepts

The term Concept has been a prominent part of investigations in psychology and neurobiology where, mostly, it is mathematically or theoretically represented. The Concepts are also studied computationally through their symbolic, distributed and hybrid representations. The majority of these approaches focused on addressing concrete concepts notion, but the view of the abstract concept is rarely explored. Moreover, most computational approaches have a predefined structure or configurations. The proposed method, Regulated Activation Network (RAN), has an evolving topology and learns representations of Abstract Concepts by exploiting the geometrical view of Concepts, without supervision. In the article, the IRIS data was used to demonstrate: the RAN's modeling; flexibility in concept identifier choice; and deep hierarchy generation. Data from IoT's Human Activity Recognition problem is used to show automatic identification of alike classes as abstract concepts. The evaluation of RAN with 8 UCI benchmarks and the comparisons with 5 Machine Learning models establishes the RANs credibility as a classifier. The classification operation also proved the RAN's hypothesis of abstract concept representation. The experiments demonstrate the RANs ability to simulate psychological processes (like concept creation and learning) and carry out effective classification irrespective of training data size.

Assembly Hierarchy Generation for Assembly System Design for a Product Family

Increased demand on product variety entails a flexible assembly system for product families which can be quickly configured and reconfigured in a responsive manner to deal with various product designs. Development of such a responsive assembly system requires an in-depth understanding of the impact of product family design on assembly system performance. In this paper, the linkage between the product family design and assembly systems is characterized by an assembly hierarchy model, which reflects a hierarchical relationship among all possible sub-assemblies and components, assembly tasks, and material flow among the tasks. Our prior research developed a recursive algorithm to generate all assembly hierarchy candidates for one single product based on its liaison graph without redundancy. These generated assembly hierarchies provide a structure to help efficiently explore optimal assembly system designs with reduced computational load. In this paper, the application of the assembly hierarchy generation algorithm will be extended to a product family by developing joint liaison graph model. Taking the advantage of the modular design of the product family, we proposed a concept of multi-level joint liaison graphs to overcome the computational challenge brought by assembly hierarchy generation for joint liaisons. Two case studies were conducted to demonstrate the algorithm.

A Parametric Network Approach for Concepts Hierarchy Generation in Text Corpus

The article presents a preflow approach for the parametric maximum flow problem, derived from the rules of constructing concepts hierarchy in text corpus. Just as generating a taxonomy can be equivalently reduced to ranking concepts within a text corpus according to a defined criterion, the proposed preflow bipush-relabel algorithm computes the maximum flow - the optimum ow that respects certain ranking constraints. The parametric preflow algorithm for generating two level concepts hierarchy in text corpus works in a parametric bipartite association network and, on each step, the maximum possible amount of ow is pushed along conditional augmenting two-arcs directed paths in the parametric residual network, for the maximum interval of the parameter values. The obtained parametric maximum ow generates concepts hierarchies (taxonomies) in text corpus for different degrees of association values described by the parameter values.