Moja lekcja

STRIPS is a method of generating plans

F

In the Forward State Propagation method, having a set of possible actions, after doing one of them, not every action from this set can be performed

T

Backward State Propagation is an example of inference in non-monotonic logic

T

In the Forward State Propagation method, we start by looking for a set of operators (actions) that in the ‘precondition’ part have clauses appearing in the description of the initial state of the planning task

F

In the ‘Triangle Table’ plans (including full descriptions of the states) are kept in mind for the potential reuse of plans or sub-plans

T

Knowledge and information are synonyms

F

Semantic knowledge is indispensable in the process of creating and working expert systems

T

Planning is inference in monotonic logic

F

A convenient representation knowledge in a planning task is a rule representation

F

The final state must be defined as conjunction of clauses without the occurrence of variables

F

The expert system usually contains both, procedural and declarative knowledge

T

Indicate (underline) which knowledge representations allow representation of inheritance: triple <O, A, W>; rule representation; semantic network; frame representation

semantic network; frame representation

Arcs in a semantic network usually connect nodes representing objects with nodes representing attributes

T

Indicate which knowledge representations are not a convenient way to represent inheritance: triple <O, A, W>, rule representation, frame representation.

triple <O, A, W>, rule representation

Indicate which knowledge representations combine the data about the represented being and the procedures for filling and modifying this data: triple <O, A, W>, rule representation, semantic network, frame representation.

frame representation.

Meta-knowledge is knowledge about knowledge, it allows to take into account exceptions included in the knowledge base

T

Each expert system is a decision support system

T

Backward Chaining can always be used in expert systems, if only data in the form of known input facts are given

F

Forward Chaining is the deductive inference (deduction)

T

In an expert system containing ‘short rules’, if we have a large set of initial facts, then the Backward Chaining method is more effective

T

List the functional modules of expert systems:

moduł pozyskiwania wiedzy moduł wnioskowania (Inference Engine) baza wiedzy (Knowledge Base) moduł objaśniająco-wyjaśniający (Explanation Facility)

Blackboard Systems consist of partial knowledge databases that are ‘specialists; within the same field, therefore their role(tasks) is the same as ‘typical’ expert systems

T

Blackboard Systems do not require any meta-knowledge needed to control the inference process

F

Blackboard Systems contain partial knowledge bases (KB) in various fields, therefore they are multi-domain expert systems

F

“Table” in Blackboard Systems is the mechanism controlling the inference process

F

In the Expert System the goal (of inference) is decomposable to a subgoals if it can be inferred based on partial subsystems (facts and knowledge)

T

In the Expert System the goal is always met if the subgoals were generated as a result of independent inference processes in a divided base

F

The Expert System using the probability theory, despite the formally correct inference, is not used in practice due to the complex calculation process

F

Certainty Factor (CF) is a heuristic method incorporated into the inference process, which causes inference to take on the properties of inference in non-monotonic logic

F

Deduction is a formally valid method of inferring the truth of a fact that it the conclusion of implication based on the truth of the premise and implications

T

Deduction and abduction are formally valid method of inference, one is used in the Forward Chaining method and the other in the Backward Chaining

F

Abduction is a formally incorrect method of inference about the potential causes of the observed fact, which is the conclusion of a true implication, on the basis of a true premise

T

Indicate which of the definitions are a formally valid method of inference (--> means implication): (p→q) is TRUE and p is TRUE so we generate correctly q is TRUE (p→q) is TRUE and q is TRUE so we generate correctly p is TRUE

(p→q) is TRUE and p is TRUE so we generate correctly q is TRUE

Inductive reasoning is just as valid as deductive, but deduction is a top-down approach, and induction - from bottom to top

F

Indicate for which type of learning we need data in the form: input and the corresponding output: supervised learning (with teacher) unsupervised learning reinforcement learning

supervised learning (with teacher)

In machine learning, we use a training set and a test set because the assumption that the sets of examples are independent of hypotheses in unfulfilled

T

In inductive learning, factors determining the choice of one hypothesis from among the set of acceptable hypotheses are called the bias of the inductive learning algorithm

T