Page 124 - Artificial Intelligence in the Age of Neural Networks and Brain Computing

1. From Aristotle’s Logic to Artificial Neural Networks and Hybrid Systems 113

Many years after Aristotle, the logic he introduced was further developed into
logic systems and rule-based systems as a foundation of knowledge-based systems
and AI. AI was perhaps pioneered by Ada Lovelace who created the first algorithm
for non-numerical data. Few types of knowledge representation and reasoning
systems are:
• Relations and implications, e.g.,: A / (implies) B.
• Propositional (true/false) logic, e.g.,: IF (A and B) or C THEN D.
• Boolean logic (George Boole).
• Predicate logic: PROLOG.
• Probabilistic logic: e.g., Bayes formula: p(A ! C)) ¼ p (C ! A). p(A)/p(C).
• Rule-based systems, expert systems, e.g., MYCIN.
Logic systems and rules, while useful, could be too rigid in some cases to represent
the uncertainty in the natural phenomena and some cognitive behavior. They are often
difficult to articulate, and in principle not adaptive to change.
Methods to complement the rule-based and logic system include fuzzy logic,
artificial neural networks, evolutionary computation, and hybrid systems as presented
further in this section. The following section presents methods and their applications
for progressing AI of evolving connectionist systems (ECOS) that allow a system
to evolve its knowledge from data through incremental learning. The last section is
presenting evolving spiking neural networks (eSNN) and their use for the creation
of brain-like AI, which is the AI that is always evolving towards human-like cognitive
behavior. Artificial Neural Networks (ANN) and SNN are the latest development in
the history of AI, and perhaps the most contemporary and promising techniques
nowadays towards building brain-inspired AI. The current development of ANN
and more specifically ECOS and SNN is linked tightly with other areas of computa-
tional intelligence, such as logic and rule based systems, evolutionary computation,
and very much with neuroscience. The future progress of ANN and SNN will be part
of the progress in the broad areas of science in general and understanding the other
related areas is essential.

1.2 FUZZY LOGIC AND FUZZY RULEeBASED SYSTEMS

Human cognitive behavior and reasoning is not always based on exact numbers
and fixed rules. In 1965, Lotfi Zadeh introduced fuzzy logic [11,12] that represents
information uncertainties and tolerance in a linguistic form. He introduced fuzzy
rules, containing fuzzy propositions and fuzzy inference.
Fuzzy propositions can have truth values between true (1) and false (0), e.g., the
proposition “washing time is short” is true to a degree of 0.8 if the time is 4.9 min,
where Short is represented as a fuzzy set with its membership function e see
Fig. 6.1.
Fuzzy rules can be used to represent human knowledge and reasoning, e.g.,
IF washing load is small
THEN washing time is short.