China
Africa
Explore chapters and articles related to this topic
Application of Artificial Intelligence Algorithms for Robot Development
Published in S. S. Nandhini, M. Karthiga, S. B. Goyal, Computational Intelligence in Robotics and Automation, 2023
R. M. Tharsanee, R. S. Soundariya, A. Saran Kumar, V. Praveen
The history of AI traces back to 1950 when Alan Turing put forth the question ‘Will machines think?’ In the year 1950, a test named the ‘Turing test’ was done to test the ability of machines to think intelligently as humans or think differently [1]. The test requires some parameters to be set in the areas of reasoning, automation, semantics and knowledge representation. As the test became successful, thereafter the term AI was introduced by John McCarthy in the year 1956. In 1970s, symbolic AI was the first evolved type of AI which uses human knowledge representation to build an AI system. After that, an agent was discovered in the 1980s which is an intelligent system which analyzes its environment and behaves correctly and intelligently. The main functions that can be performed by AI systems include planning, organizing, decision making and task execution. Many scientists predicted that the customer request can be processed by machines without human intervention by using AI techniques. AI is developing every day. The applications of AI are increasing rapidly in a variety of fields like medicine, Robotics, automation and so on [2].
Even More AI!
Published in Nicolas Sabouret, Lizete De Assis, Understanding Artificial Intelligence, 2020
In symbolic AI, machines attempt to reason about the knowledge contained in data, just like we do. For this, knowledge representation models are developed to program computers with the knowledge they are meant to manipulate. The first general knowledge representation model was proposed by Richard Richens in 1956. He dubbed this model a semantic network, and it led to many other advancements in artificial intelligence such as John Sowa’s conceptual graphs, proposed in 1976, and Ronald Brachman’s description logics,3 proposed in the mid 1980s.
Artificial Intelligence and Machine Learning for the Industrial Internet of Things (IIoT)
Published in Anand Sharma, Sunil Kumar Jangir, Manish Kumar, Dilip Kumar Choubey, Tarun Shrivastava, S. Balamurugan, Industrial Internet of Things, 2022
Approaches in the AI field are primarily categorized into two. The first one being the traditional symbolic AI, which has been leading throughout the history, is categorized according to an abstraction that is of high level and a perceptual view that is macroscopic. Systems that involve knowledge engineering and logic programming are categorized under this category. The areas of logical reasoning, systems based on knowledge, symbolic ML, natural language processing, and search techniques fall under symbolic AI.
A holonic architecture for the supply chain performance in industry 4.0 context
Published in International Journal of Logistics Research and Applications, 2021
Kamar Zekhnini, Anass Cherrafi, Imane Bouhaddou, Abla Chaouni Benabdellah, Rakesh Raut
Artificial intelligence (AI): Since the late ‘ seventies, To develop and construct ‘thinking machines’ capable of imitating, learning, and replacing human intelligence, artificial intelligence (AI) was introduced (Min 2010). AI, known in many branches as soft computing, refers to advanced machine learning, expertise, and decision-making technologies. AI utilises consciousness and perception in contrast to analytical methods (Mohammadi and Minaei 2019). The particular AI techniques being utilised can range from traditional symbolic AI, relying on mathematical or knowledge-based representations of problems, to sub-symbolic AI, including, for example, fuzzy systems and evolutionary computation, to statistical AI, encompassing approaches to machine learning (Baryannis et al. 2018).
Advanced design of complex façades
Published in Intelligent Buildings International, 2018
Winfried Heusler, Ksenija Kadija
Artificial Intelligence (AI) tries to imitate human intelligence. ‘Strong Artificial Intelligence’ claims to reach or even surpass people in all of the above categories. It already succeeds in doing this today in logical thinking, e.g. in making decisions and in planning, learning and communicating in natural language. Emotional and social AI, on the other hand, are still fiction. ‘Weak Artificial Intelligence’ does not reach this high standard. It is designed to perform clearly defined tasks and is already used in everyday life, for example in expert and navigation systems. It is also behind speech recognition and intelligent search algorithms. There are two alternative approaches (Jeschke 2014): The symbolic AI relies on existing knowledge and logic.The neuronal AI acquires knowledge through experience and is able to learn independently.
Monte Carlo science
Published in Journal of Turbulence, 2020
There are several distinctions to be made before we begin our argument. The first one has to do with how AI is supposed to work [3], which is traditionally divided into symbolic [4] and sub-symbolic [5]. Symbolic AI is the classical kind, which manipulates symbols representing real-world variables according to rules set by the programmer, typically embodied into an ‘expert system’. For example, assuming some personal definition of vortices, AI can be used to encapsulate this knowledge into rules to identify and isolate them. On the contrary, sub-symbolic AI is not interested in rules, but in algorithms to do things. For example, given enough snapshots of a flow in which vortices have been identified (maybe by a pre-existing expert system), we can train a neural network to distinguish them from vorticity sheets. The result of sub-symbolic AI is not the rule, but the algorithm, and it does not imply that a rule exists. After training our system, we may not know (or care) what a vortex is or what distinguishes it from a sheet, but we may have a faster way of distinguishing one from the other than what would have been possible using only pre-ordained physics-based rules. Copernicus and the Greeks had symbolic representations of the day-night cycle, although with very different ideas of the rules involved. Most other living beings, which can distinguish night from day and usually predict quite accurately dawn and dusk, are (probably) sub-symbolic. The classical scientific method, including causality, is firmly on the symbolic side of the divide: we are not only interested in the result but also in the rule. But there is a small but growing body of scientists and engineers, who feel that data and a properly trained algorithm are all the information required about nature, and that no further rule is necessary, as for example discussed in Refs. [6,7].