Artificial neural networks, inspired by the human brain, are a sub-area of machine learning as the backbone of modern deep learning algorithms. Well-known variants of artificial neural networks include support vector machine, radial basis network, recurrent neural network, auto encoder, Boltzmann machine, and deep convolutional network. These models have been used in a variety of tasks such as predictive modeling, data processing, and adaptive control for their impressive performance. However, one major drawback of artificial neural networks lies in its opaque nature with inferences hardly making sense to human reasoning, thus generally being considered as the black-box modeling approach.
Inspired by how the information is perceived and processed by human users, granular computing is an emerging computing paradigm that concerns the processing of complex information entities called "information granules". Considering the similarity, functional adjacency, indistinguishability, or coherency of numeric entities, the information granules may be established within various settings including set theory, interval calculus, fuzzy sets, rough sets, shadowed sets, and probabilistic granules. Generally, since granular computing is about the formation, processing, and communication of these information granules, it devotes itself to building more robust and transparent models or constructs that facilitate the understanding of the extracted knowledge.
Obviously, artificial neural networks bring a wealth of learning abilities to the built intelligent systems (e.g., robotics and cyber-physical systems) being realized both in supervised and unsupervised modes. Granular computing, on the other hand, comes with a more human-centric manner of knowledge representation and information processing. Hence, the hybridization of artificial neural networks and granular computing is potentially complementing their mutual strengths. This thus inspires to develop the granular computing-based artificial neural networks in which learning comes hand in hand with the robustness and transparency of the resulting structure. By combining both artificial neural networks and granular computing, this Research Topic aims to contribute to the development of intelligent systems and promote that the newly built neural network models are robust, transparent, and trusty when being applied to robotics, cyber-physical systems, and embodiment of such systems in software and/or hardware devices.
We welcome research contributions (both original research and comprehensive survey articles) from all related areas, with a focus on, but not limited to, the following topics:
• Interval-based artificial neural networks
• Fuzzy set-based artificial neural networks
• Type-2 fuzzy set-based artificial neural networks
• Rough set-based artificial neural networks
• Shadow set-based artificial neural networks
• Probabilistic granules-based artificial neural networks
• Autonomous robotics with granular computing-based artificial neural networks
• Advanced control algorithm for special-type robots
• Embodiment and human-centered design
Artificial neural networks, inspired by the human brain, are a sub-area of machine learning as the backbone of modern deep learning algorithms. Well-known variants of artificial neural networks include support vector machine, radial basis network, recurrent neural network, auto encoder, Boltzmann machine, and deep convolutional network. These models have been used in a variety of tasks such as predictive modeling, data processing, and adaptive control for their impressive performance. However, one major drawback of artificial neural networks lies in its opaque nature with inferences hardly making sense to human reasoning, thus generally being considered as the black-box modeling approach.
Inspired by how the information is perceived and processed by human users, granular computing is an emerging computing paradigm that concerns the processing of complex information entities called "information granules". Considering the similarity, functional adjacency, indistinguishability, or coherency of numeric entities, the information granules may be established within various settings including set theory, interval calculus, fuzzy sets, rough sets, shadowed sets, and probabilistic granules. Generally, since granular computing is about the formation, processing, and communication of these information granules, it devotes itself to building more robust and transparent models or constructs that facilitate the understanding of the extracted knowledge.
Obviously, artificial neural networks bring a wealth of learning abilities to the built intelligent systems (e.g., robotics and cyber-physical systems) being realized both in supervised and unsupervised modes. Granular computing, on the other hand, comes with a more human-centric manner of knowledge representation and information processing. Hence, the hybridization of artificial neural networks and granular computing is potentially complementing their mutual strengths. This thus inspires to develop the granular computing-based artificial neural networks in which learning comes hand in hand with the robustness and transparency of the resulting structure. By combining both artificial neural networks and granular computing, this Research Topic aims to contribute to the development of intelligent systems and promote that the newly built neural network models are robust, transparent, and trusty when being applied to robotics, cyber-physical systems, and embodiment of such systems in software and/or hardware devices.
We welcome research contributions (both original research and comprehensive survey articles) from all related areas, with a focus on, but not limited to, the following topics:
• Interval-based artificial neural networks
• Fuzzy set-based artificial neural networks
• Type-2 fuzzy set-based artificial neural networks
• Rough set-based artificial neural networks
• Shadow set-based artificial neural networks
• Probabilistic granules-based artificial neural networks
• Autonomous robotics with granular computing-based artificial neural networks
• Advanced control algorithm for special-type robots
• Embodiment and human-centered design