Towards integrated neural–symbolic systems for human-level AI: Two research programs helping to bridge the gaps

Abstract

This is the accepted version of the paper.This version of the publication may differ from the final published version. signals, but cognition also exhibits the capability to efficiently perform abstract reasoning and symbol processing; in fact, processes of the latter type seem to form the conceptual cornerstones for thinking, decision-making, and other (also directly behavior-relevant) mental activities (see, e.g., Fodor & Pylyshyn (1988)). (2015)). This has several reasons also beyond the 35 analogy to the described functioning principles of the brain: Permanent repository link• In general, network-based approaches possess a higher degree of biological motivation than symbol-based approaches, also outmatching the latter in

Modeling Cognitive Functionalities of Prosthetic Arm using Conceptual Spaces

“…the relationship between the attributes best convey the relationship between the concepts [8]. It should be also noted that concepts of cognitive frameworks models real world instance based on their static properties (attributes).…”

“…A concept in the conceptual space is a pair namely configuration (object) and description of the configuration (attribute). In this work, we use three granules of attribute description for each configuration namely focal attribute, general attribute and essential attribute [8]. This proposal regards an attribute as a focal attribute if the attribute is possessed by all the configurations (cardinality of focal attribute set is greater than 2).…”

“…Generally, a concept is a pair consisting of a real world entity and its description. However, literature suggests that the conventional notion of concept does not reflect the hierarchical structure in human brain [8]. On the other hand, Gӓ� rdenfors suggest that notion of concept as 'regions' in conceptual space framework lacks precision without topological structure [9].…”

“…One straight forward approach is to classify the arbitrary information into information granules such as necessary, sufficient and necessary as well as sufficient granules [26]. Similarly, concepts can also be learned using information granules [8]. Alternatively, concepts can be learned from incomplete information via granular computing techniques [27].…”

A dynamical-structure neural network model specified for representing logical relations with inhibitory links and fewer neurons

“…The reason is the functional-approximation based ANN models, aimed for representing numeric relations, are not suitable for representing and storing logical relations in their neural network structures according to the "no free lunch" theorem [16], [17]. Aiming to make the neural network have the ability to deal with logical issues, researchers have been trying to design novel neural network models from two aspects: neural components and neural connecting styles to make neural network model can represent and store logical relations [18], [19]. Through the research, they want to combine symbolic logic with ANN which are yet developing individually, and they want to apply ANN into more application areas like knowledge representation and reasoning, expert systems, semantic web and cognitive modelling and robotics.…”