Unit 40: Knowledge-based Systems and Techniques


Unit 40:

Knowledge-based Systems and

 

 

Techniques


Unit code:
A/601/1446

QCF level:
5


Credit value:
15






Aim

This unit will introduce learners to the concepts and techniques used in artificial intelligence and knowledge-based systems and develop an understanding of rule-based systems, fuzzy logic and artificial neural networks.

Unit abstract

The unit starts by introducing learners to knowledge bases and rule bases that are used extensively in expert systems, and at a much lower level are used for simple reasoning/logic operations. The concept of rule bases is extended to fuzzy operations and fuzzy logic which is increasingly being used in domestic appliances and is in use in many industrial applications. Finally, learners are introduced to artificial neural networks, which are related to basic brain (synapse) functions, and ‘learning’ is demonstrated using simple neuron structures. Evaluation of fuzzy logic algorithms and artificial neural networks is achieved via simulation using proprietary software.

Learning outcomes

On successful completion of this unit a learner will:

1     Understand the use of knowledge-based and rule-based systems

2       Be able to use fuzzy logic

3       Be able to use artificial neural networks.




Unit content
Understand the use of knowledge-based and rule-based systems

Knowledge and rule base: terminology (facts and rules, propositions or predicates, deep and surface knowledge – heuristics); semantic networks; forward chaining; antecedents and consequences; conflict resolution; backward chaining; applications and implementation (identification of examples where such systems would be used)


2      Be able to use fuzzy logic

Human analogy: human reasoning and expert knowledge

Fuzzy logic theory: conventional binary logic; crisp and fuzzy sets; fuzzy reasoning; fuzzy rules; membership functions; inference engines; de-fuzzification

Applications: identification and analysis of examples eg cameras, domestic appliances, industrial equipment and processes

Implementation: development of fuzzy rules; evaluation of performance via simulation


3      Be able to use artificial neural networks

Biological analogy: synapse, axons, dendrites

Network topologies and operating characteristics: Hopfield networks; multi-layer perceptron; back propagation; self organising networks; Kohonen networks; radial basis function networks; neuro-fuzzy and fuzzy-neural

Applications: identification and analysis of examples eg pattern classification, optical character recognition, image analysis, biometrics

Implementation: experimentation with neural network configurations; learning coefficients: RMS; error evaluation of performance via simulation



Learning outcomes and assessment criteria


Learning outcomes
Assessment criteria for pass


On successful completion of
The learner can:


this unit a learner will:










LO1 Understand the use of

1.1
explain knowledge-base and rule-base terminology


knowledge-based and rule-

1.2
devise and interpret semantic networks


based systems










1.3
describe applications of knowledge-based and rule-





based systems








LO2 Be able to use fuzzy logic

2.1
describe human reasoning and expert knowledge




2.2
use fuzzy logic theory to produce fuzzy rules,





fuzzification and defuzzification




2.3
describe and evaluate applications of fuzzy logic




2.4
design and evaluate fuzzy logic systems using





appropriate software







LO3 Be able to use artificial

3.1
explain the biological analogy of neural networks


neural networks

3.2
explain network topologies and operating characteristics








3.3
describe and evaluate applications of neural networks




3.4
design and evaluate neural networks using appropriate





software.







Guidance
Links

This is a stand-alone unit.


Essential requirements

The use of software packages is an essential part of the delivery of this unit. Proprietary software such as MATLAB/Simulink, or equivalent, with appropriate tool boxes for fuzzy logic and neural networks must be available to learners.

Employer engagement and vocational contexts

The delivery of this unit will benefit from centres establishing strong links with employers willing to contribute to the delivery of teaching, work-based placements and/or detailed case study materials.

Comments