1 of 28
Definition
domain expert: knowledgeable and skilled person capable of
solving problems in a specific person capable of solving problems
in a specific area or domain
knowledge engineer: someone who is capable of designing,
building and testing an expert system. He or she interviews the
domain expert to find out how a particular problem is solved.
programmer: the person responsible for the actual programming,
describing the domain knowledge in terms that a computer can
understand.
project manager: leader of the expert system project, responsible
for keeping the project on track
end-user: the user, must be comfortable using it
Give this one a try later!
Leg uit wat wordt bedoeld met een What is the difference between
"epoch" in de context van ANN's. classical and fuzzy rules
, Wat zijn de 5 spelers in het
Wat wordt bedoeld met "deep" in
ontwikkelteam van een expert
de zin van "deep neural nets"?
systeem?
Don't know?
2 of 28
Term
Bij expertsystemen gebaseerd op fuzzy logic worden verschillende
technieken gebruikt om een conclusie te genereren. Twee van
dergelijke technieken zijn de Mamdani en Sogena stijl afleiding. Wat
is het verschil tussen Mamdani en Sugeno inference stijl?
Give this one a try later!
1. knowledge base
2. database
3. inference engine
4. explanation facilities
5. User interface
Mamdani
• Intuitive
• Well-suited to human input
• More interpretable rule base
• Have widespread acceptance
Sugeno
• Computationally efficient
• Work well with linear techniques, such as PID control
• Work well with optimization and adaptive techniques
, • Guarantee output surface continuity
• Well-suited to mathematical analysis
With supervised learning, a ANN has inputs and outputs, you train the network
using data where you have the inputs and outputs, that way you can validate the
network, tweak the weights and see how accurate it is. With unsupervised
learning, the ANN only has an input and no teacher or a way to know the right
answer.
Image recognition is supervised learning.
Relations between rules: makes it difficult to observe how individual rules serve
the overal strategy
Search strategy: Systems with a large set of rules (like, >100) can be slow
Inability to learn: cannot modify its knowledge base automatically. The
knowledge engineer (a person) is still responsible for revising the system
Don't know?
3 of 28
Term
A neural network can be described as comprised of three layer
types. Which layer type is associated with the data to be analyzed?
Give this one a try later!
supervised learning input layer
unsupervised learning output layer
Don't know?
Definition
domain expert: knowledgeable and skilled person capable of
solving problems in a specific person capable of solving problems
in a specific area or domain
knowledge engineer: someone who is capable of designing,
building and testing an expert system. He or she interviews the
domain expert to find out how a particular problem is solved.
programmer: the person responsible for the actual programming,
describing the domain knowledge in terms that a computer can
understand.
project manager: leader of the expert system project, responsible
for keeping the project on track
end-user: the user, must be comfortable using it
Give this one a try later!
Leg uit wat wordt bedoeld met een What is the difference between
"epoch" in de context van ANN's. classical and fuzzy rules
, Wat zijn de 5 spelers in het
Wat wordt bedoeld met "deep" in
ontwikkelteam van een expert
de zin van "deep neural nets"?
systeem?
Don't know?
2 of 28
Term
Bij expertsystemen gebaseerd op fuzzy logic worden verschillende
technieken gebruikt om een conclusie te genereren. Twee van
dergelijke technieken zijn de Mamdani en Sogena stijl afleiding. Wat
is het verschil tussen Mamdani en Sugeno inference stijl?
Give this one a try later!
1. knowledge base
2. database
3. inference engine
4. explanation facilities
5. User interface
Mamdani
• Intuitive
• Well-suited to human input
• More interpretable rule base
• Have widespread acceptance
Sugeno
• Computationally efficient
• Work well with linear techniques, such as PID control
• Work well with optimization and adaptive techniques
, • Guarantee output surface continuity
• Well-suited to mathematical analysis
With supervised learning, a ANN has inputs and outputs, you train the network
using data where you have the inputs and outputs, that way you can validate the
network, tweak the weights and see how accurate it is. With unsupervised
learning, the ANN only has an input and no teacher or a way to know the right
answer.
Image recognition is supervised learning.
Relations between rules: makes it difficult to observe how individual rules serve
the overal strategy
Search strategy: Systems with a large set of rules (like, >100) can be slow
Inability to learn: cannot modify its knowledge base automatically. The
knowledge engineer (a person) is still responsible for revising the system
Don't know?
3 of 28
Term
A neural network can be described as comprised of three layer
types. Which layer type is associated with the data to be analyzed?
Give this one a try later!
supervised learning input layer
unsupervised learning output layer
Don't know?
