Term
|
Definition
Turing Test Approach
Cognitive Modeling Approach
Laws of Rational Thought Approach
Rational Agent Approach |
|
|
Term
| Main Features of the Cognitive Modeling Approach |
|
Definition
| Attempted to create an agent modelled on the human brain. Based in psychology |
|
|
Term
Classify each approach according to whether they are based on thought or action and on human standards or ideal standards and justify (Cognitive Modeling, Turing test, Rational Agent and Laws of Rational Thought) |
|
Definition
Cognitive -Thinking human
Turing -Acting human
Rational Thought -Thinking rationally
Rational Agent - Acting rationally |
|
|
Term
| Discuss the relative advantages/disadvantages of the Turing Test approach |
|
Definition
- Pro: Empirical, does not define intelligence
- Con: Limited to symbolic tasks, human based, cannot have original thought, cannot create perfect rule-set
|
|
|
Term
| What is meant by a rational agent? |
|
Definition
| An agent that behaves optimally |
|
|
Term
| Does a rational agent always do the right thing? Explain. |
|
Definition
| Yes, as it performs the optimal action based on a set of given beliefs and goals. |
|
|
Term
| What is meant by a percept? |
|
Definition
| One unit of sensory input tracked by an agent about its environment |
|
|
Term
| Explain classifying an environment by fully observable v partially observable |
|
Definition
| Fully observable environment is one in which agent can perceive its complete state. |
|
|
Term
| Explain classifying an environment by single v multiagent |
|
Definition
- Environment is multiagent if performance measures depend on other agent's behavior. If an environment is multiagent, it is classified cooperative or competitive
|
|
|
Term
| Explain classifying an environment by deterministic v stochastic |
|
Definition
| In deterministic, next action of agent determined solely by current state and action of the agent. Stochastic (aka nondeterministic) environments are those where past actions/conditions/states impact current behavior. |
|
|
Term
| Explain classifying an environment by episodic v sequential |
|
Definition
| Episodes are single actions independent of each other based on a single percept. Episodic environments are where the results of one episode don't effect other episodes, and thus an agent does not need to plan ahead wrt episodes. Sequential is when one action may influence future actions. |
|
|
Term
| Explain classifying an environment by static v dynamic |
|
Definition
| Static environments are only changed by an agent's actions. Semidynamic environment is one that is static but where the performance measure is time-dependent |
|
|
Term
| Explain classifying an environment by discrete v continuous |
|
Definition
| Discrete environments have a discrete number of combinations of environment states, actions and percepts. |
|
|
Term
| Explain classifying an environment by known v unknown |
|
Definition
| Unknown environments are those where the agent does not know the rules of play. Unknown does not always mean partially observable. |
|
|
Term
| What are the 5 types of agents as identified by Russell and Norvig? |
|
Definition
-Table Lookup
-Simple Reflex
-Reflex with memory
-Goal-based
-Utility-based |
|
|
Term
| Describe a Table Lookup agent and explain its advantages and disadvantages. |
|
Definition
- Table Look-up - Look through table for option based on table of percepts
- Pro: Simplest, can be used with other agent types
- Con: Size of table, construction of table, not autonomous
|
|
|
Term
| Describe contributions to AI from external fields |
|
Definition
Math - logic, numerical representations of conditions
Linguistics - Symbols, how data is represented/interpretted
Psych - Understanding of human mind, thoughts, processes
Philosophy - the mind body problem |
|
|
Term
| What is meant by an 'autonomous agent'? |
|
Definition
| An agent that can act under its own control |
|
|
Term
| What is meant by 'utility'? |
|
Definition
| The measure of the desirability of a state |
|
|
Term
| What is meant by a 'performance measure'? |
|
Definition
| A criterion for success, used to evaluate environmental states that result from an agents actions. Should be objective but can be qualitative or quantitative. |
|
|
Term
|
Definition
| (defun fname (param-list body)) |
|
|
Term
|
Definition
| atom(INPUT) returns non-NIL if INPUT is an atomic entry |
|
|
Term
|
Definition
(cond (test s-expression-sequence)
(test s-expression-sequence) ...
)
|
|
|
Term
|
Definition
(cdr dotted-pair) -cdr stands for "contents of decrement register" -Returns right pointer of dotted pair |
|
|
Term
|
Definition
(car dotted-pair) -car stands for "contents of address register" -Returns left pointer of dotted pair |
|
|
Term
|
Definition
(set s1 s2 ... ) -Binds value of s2 to s1, ... -Returns rst value bound |
|
|
Term
|
Definition
| (setq s1 s2) (set (quote s1) s2) |
|
|
Term
| LISP - Compare varieties of 'equal' |
|
Definition
- (eq s1 s2): T if s1 and s2 point to the same object
- (eql s1 s2): T if s1 and s2 point to the same object, or if same number (and same type), or same characters
- (equal s1 s2): T if s1 and s2 evaluate to same value
- (equalp s1 s2): Same as equal, but returns T if s1 and s2 are same number regardless of type
|
|
|
Term
|
Definition
- (map-fn-name applied-fn lists)
- lists must have same numbers of elements
- Must be 1 list for each argument of applied-fn
|
|
|
Term
|
Definition
- (cons a b)
- Cons cell (dotted pair) - More primitive than list
- Syntax: (lobject . robject)
- Consists of a "left" pointer lobject and a "right" pointer robject (to S-expressions
|
|
|
Term
| Internal representation of cons cells and lists |
|
Definition
|
|
Term
| Discuss the relative advantages/disadvantages of the Cognitive Modelling approach |
|
Definition
- Pro: Thinks like a human, can be conducted through multiple avenues (introspection, experiments, brain imaging)
- Con: Requires understanding of how people think
|
|
|
Term
| Discuss the relative advantages/disadvantages of the rational agent approach |
|
Definition
- Pro: More general than Law of Thought, more applicable to scientific analysis
- Con: Finding a perfect/best COA is not always possible, and is impossible in any complex environment
|
|
|
Term
| Discuss the relative advantages/disadvantages of the Laws of Rational Thought approach |
|
Definition
- Pro: Formal, provable, mathematical approach
- Con: difficult language to use, limited computational resources
|
|
|
Term
|
Definition
Performance
Environment
Actuators
Sensors |
|
|
Term
| What are the 5 components that define a problem? Describe each. |
|
Definition
- Initial State - State from which problem solving begins
- Action - a step the agent can carry out
- Transition model - results of the action
- Goal test - tests to see if it is the goal state
- Path cost function - computes cost of path from initial state to current state
|
|
|
Term
| What is a 'multistate problem'? Describe the conditions that lead to this kind of problem. |
|
Definition
| When an agent has limited knowledge of the results of its actions or the state it is in. This can result when actions don't have the desired results |
|
|
Term
| What is the difference between and implicit and explicit search graph? What are the pros and cons of each? What determines which is used? |
|
Definition
| Explicit generates the entire tree and stores it, implicit dynamically generates the tree. Explicit can be result of previous searches, but is memory and computationally expensive. Implicit is more efficient in space and time, and only maintains nodes that have been visited. |
|
|
Term
| List several factors that influence whether to perform goal-directed or data-directed search, and what their influence is. |
|
Definition
- Branching factor: move in direction of smaller branching factor
- Are actions reversible: cannot do goal-directed if not
- Number of targets: move in direction that has greater number of targets
|
|
|
Term
| What are the ramifications of tree v graph representations for search algorithms? |
|
Definition
| Trees allow infinite paths and have large storage requirements, making graphs preferable. |
|
|
Term
| What distinguishes a local search algorithm from other types? |
|
Definition
| Local looks in the immediate area or notably limits the number of states checked to find the closest solution or closest best solution |
|
|
Term
| LEARN HOW TO TRAVERSE TREE IN DFS, BFS, steepest-ascent hill-climbing, and A* |
|
Definition
|
|
Term
| Explain how uniform cost and BFS are similar. Explain how they are different. |
|
Definition
| UCS is like BFS but finds cheapest path to goal by selecting a node in the fringe with the cheapest cost. UCS uses a priority queue. |
|
|
Term
| What distinguishes recursive best first search from A*? |
|
Definition
| RBFS assigns a back up value to each node, and it only maintains the most promising path to the goal at any given time. |
|
|
Term
| Briefly describe how simulated annealing search works. |
|
Definition
| Simulated annealing works by always selecting random moves, and then committing if the move improves the situation. If the move is not desirable, it will be selected based on a mathematical formula. |
|
|
Term
| Briefly describe how genetic search algorithm works. |
|
Definition
- Start with k generated states
- Spawn a new generation
- Choose best chromosomes of the new generation
- Randomly combine pairs of good chromosomes
- Continue creation and combination of chromosomes until solution or some pre-set limit reached.
|
|
|
Term
| What does it mean for a heuristic to be 'admissible'? What are the ramifications for A*? |
|
Definition
| An 'admissible' heuristic is one that never over-estimates the cost from n to g. A* is not optimal if the heurisitic is not admissible. |
|
|
Term
| What is 'dominance' and how does it influence the choice of a heuristic? |
|
Definition
ha dominates hb if ha(n) ≥ hb(n)
Dominating heuristic will expand fewer nodes and be more accurate, making it the preferred choice. |
|
|
Term
| How does 'problem relaxation' relate to heuristic function creation? Give an example. |
|
Definition
A relaxed problem is one with fewer constraints, and the cost of the optimal solution to the relaxed problem is an admissible heuristic in the original.
EXAMPLE: The 8 puzzle |
|
|
Term
Is the following algorithm optimal? Is it complete? If not optimal or not complete, explain. Breadth First Search |
|
Definition
|
|
Term
Is the following algorithm optimal? Is it complete? If not optimal or not complete, explain. Depth First Search |
|
Definition
| Complete if a graph, not complete if a tree. Not optimal because it may find a solution at the end of a path that is more expensive than another solution that it hasn't reached yet. |
|
|
Term
Is the following algorithm optimal? Is it complete? If not optimal or not complete, explain. Uniform Cost Search |
|
Definition
|
|
Term
Is the following algorithm optimal? Is it complete? If not optimal or not complete, explain. Depth Limited Search |
|
Definition
| Not complete because there may be a solution beyond it's limited depth. Optimal iff the limit is deep enough. |
|
|
Term
Is the following algorithm optimal? Is it complete? If not optimal or not complete, explain. Iterative Deepening Depth First Search |
|
Definition
|
|
Term
Is the following algorithm optimal? Is it complete? If not optimal or not complete, explain. Best First Search |
|
Definition
| Complete for graphs, not for nodes. Not optimal as it is only concerned with what looks like the best from a given node. |
|
|
Term
Is the following algorithm optimal? Is it complete? If not optimal or not complete, explain. A* Search |
|
Definition
| Optimal (if heuristic is admissible) and Complete |
|
|
Term
Is the following algorithm optimal? Is it complete? If not optimal or not complete, explain. Greedy Search |
|
Definition
| Not optimal as the algorithm often commits to a path based on local values. Complete if a graph, not complete if a tree. |
|
|
Term
Is the following algorithm optimal? Is it complete? If not optimal or not complete, explain. Genetic Algorithm Search |
|
Definition
| Not optimal or complete due to inherent randomness. |
|
|
Term
Is the following algorithm optimal? Is it complete? If not optimal or not complete, explain. Steepest-Ascent Hill Climbing Search |
|
Definition
| Not optimal as it only looks at local values, and not complete as it can get stuck at local minima. |
|
|
Term
| List several desirable characteristics of a knowledge representation scheme/language. |
|
Definition
- Concise
- Unambiguous
- Can add to it
|
|
|
Term
| What does it mean for an inference engine to be 'sound'? |
|
Definition
| It does not create spurious inferences. |
|
|
Term
| What does it mean for an inference engine to be 'complete'? |
|
Definition
| It can/does create/check all possible inferences. |
|
|
Term
| What is an 'interpretation' with respect to propositional logic? |
|
Definition
| The mapping of the value of a propositional sentence to true or false. |
|
|
Term
| What is meant by a 'model' of a set of propositional sentences? |
|
Definition
| A world where all the propositional sentences in the set are true. |
|
|
Term
| What does it mean for a propositional sentence to be 'satisfiable'? |
|
Definition
| It is true in at least one model. |
|
|
Term
| Using model checking, show that P=>Q Λ ¬Q |= ¬P |
|
Definition
|
|
Term
| RESOLUTION IN PROPOSITIONAL LOGIC |
|
Definition
|
|
Term
| Explain several problems associated with propositional logic as a knowledge representation scheme for real world problems. |
|
Definition
- Locational Info - Difficult to represent that something is only in one place at a time
- Relational Info - many real world relations preclude others (facing east means you can't be facing west)
- Actions - actions must have sufficient preconditions to prevent accidental simultaneous execution
|
|
|
Term
| What 3 things are mapped to a model by an interpretation? |
|
Definition
- Constants to objects
- Predicates to relations among objects
- Functions to functions in the domain
|
|
|
Term
| Explain why Ãx P(x) => Q(x) is valid in all models. Ã represents the universally qualifier. |
|
Definition
If an object is substituted for x that makes P(x) false, that particular implication will be true; if the object makes P(x) true, then Q(x) is true (by modus ponens) and the implication is true again. |
|
|
Term
What are the semantics of
UNIVERSALLY QUALIFIED(x) P(x) |
|
Definition
| The semantics are that the sentence is true if P(x) is true for all x in the domain. |
|
|
Term
| Characteristics of Database semantics |
|
Definition
- Closed world assumption
- Closed domain assumption
- Unique names assumption
|
|
|
Term
| What are the semantics of F(x) where F is a function? |
|
Definition
| Returns an object in the domain. |
|
|
Term
| Steps to convert from FOL to CNF |
|
Definition
- Eliminate implication
- Reduce scope of NOT
- Standardize variables
- Eliminate existential
- Move quantifiers to left
- Drop prefix
- Convert into conjunctions of disjuncts
- Write as a set of clauses
- Rename variables so each clause has unique variables
|
|
|
Term
| Why isn't propositionalization a good approach to proofs in FOL? |
|
Definition
| KB becomes very large. FOL uses far fewer axioms because it uses general rules instead of a proposition for each and every individual case. |
|
|
Term
| Discuss 2 ways to make retrieval more efficient during unification. |
|
Definition
- Using more efficient data structures: Linked list, predicate indexing, multi-indexing in order from worst to best
- Using a more efficient search strategy for ASK
|
|
|
Term
| Describe a Simple Reflex agent and explain its advantages and disadvantages. |
|
Definition
- Simple Reflex - exact cause and effect
- Pro: Simple, predictable
- Con: No internal representation of world, prone to infinite loops
|
|
|
Term
| Describe a Reflex w/Memory agent and explain its advantages and disadvantages. |
|
Definition
- Reflex w/Memory - reflex based with knowledge of world
- Pro: utilizes knowledge, uses rules
- Con: No mechanism for making decisions
|
|
|
Term
| Describe a Goal-Based agent and explain its advantages and disadvantages. |
|
Definition
- Goal-based - incorporation of goals allows for decision making
- Pro: decision making, plans, more flexible than reflex based agents
- Con: slower than reflex as it must reason
|
|
|
Term
| Describe a Utility-Based agent and explain its advantages and disadvantages. |
|
Definition
- Utility-based - goal based with a a heurisitic about which decisions to choose
- Pro: chooses desirable way to goal
- Con: slowest, most complex
|
|
|
Term
| Main features of the Turing Test Approach |
|
Definition
Describe experiment (1 AI and 1 human are asked questions by a human interrogator, and AI completes test if interrogator is unable to determine responses are from a human) |
|
|
Term
| Main Features of the Rational Agents Approach |
|
Definition
Rational Agent - Agent acts intelligently, given a set of beliefs and goals, agent performs appropriate actions to achieve those goals |
|
|
Term
| Main Features of the Laws of Rational Thought Approach |
|
Definition
Laws of Rational Thought - Thinking rationally, based in math and seeks the best answer. may run forever if no solution exists |
|
|