Hybrids

Question 1

How can learning be applied to a neuro-fuzzy system?

Answer 1

A neuro-fuzzy system is essentially a multi-layer neural network, and thus it can apply standard learning algorithms developed for neural networks, including the back-propagation algorithm:

When a training input-output example is presented to the system, the back-propagation algorithm computes the system output and compares it with the desired output of the training example. The difference (also called the error) is propagated backwards through the network from the output layer to the input layer. The neuron activation functions are modified as the error is propagated. To determine the necessary modifications, the back- propagation algorithm differentiates the activation functions of the neurons.

Question 2

What neuro-fuzzy system parameters can be learned during training?

Answer 2

• weights
• fuzzy rules

Question 3

Discuss two alternative ways to combine neural networks and fuzzy systems to form a neurofuzzy system

Answer 3

• Approach 1:
  
  • NN and FS work independently of each other.
  • The combination lies on the determination of certain parameters of a FS by a NN or a NN learning algorithm
• Approach 2:
  
  • define a homogeneous architecture, usually like the structure of a NN.
  • This can be done by interpreting the FS as a special kind of a NN, or by implementing a FS using a NN.

Question 4

Why are neural networks and fuzzy systems combined?

Answer 4

• The most important reason for combining fuzzy systems and neural networks is their learning capability.
• This kind of combination should be able to learn linguistic rules and/or membership functions, or to optimise existing ones.

Common way to apply a learning algorithm to a fuzzy system

1. Represent the fuzzy system in a special neural-network- like architecture
2. Use a learning algorithm to train the system

• i. Replace not differentiable functions with differentiable
• ii. Do not use gradient-based learning but a better suited procedure

Question 5

Discuss the advantages and disadvantages of NNs and Fuzzy systems and explain why combining them into a hybrid intelligent system makes sense.

Answer 5

The most important reason for combining fuzzy systems and neural networks is their learning capability. This kind of combination should be able to learn linguistic rules and/or membership functions, or to optimise existing ones.

Question 6

What is approximate reasoning?

Answer 6

• In a rule-based expert system, the inference engine compares the condition part of each rule with data given in the database. When the IF part of the rule matches the data in the database, the rule is fired and its THEN part is executed.
• In rule-based expert systems, the precise matching is required. As a result, the inference engine cannot cope with noisy or incomplete data.
• Neural expert systems use a trained neural network in place of the knowledge base. The neural network is capable of generalisation.
• In other words, the new input data does not have to precisely match the data that was used in network training. T
• his allows neural expert systems to deal with noisy and incomplete data.
• This ability is called approximate reasoning.
• The rule extraction unit examines the neural knowledge base and produces the rules implicitly ‘buried’ in the trained neural network.
• The explanation facilities explain to the user how the neural expert system arrives at a particular solution when working with the new input data.
• The user interface provides the means of communication between the user and the neural expert system.