Your Answer : (Not Answered)

Correct Answer : D

The four Rs are retrieve a similar case, reuse its solution, revise it against the outcome, and retain the revised case for future use. Retention is what makes the system improve with experience. The acquisition cycle, the production system cycle and plan-generate-test belong elsewhere in this topic. The mnemonic is standard and is asked for verbatim often enough to be worth learning as such. A limit that survives making every external component ideal is an intrinsic one, and worth knowing for that reason.

Your Answer : (Not Answered)

Correct Answer : B

Sensing is worth performing only if what is learned would change what the agent does, and the value is the resulting improvement in expected outcome. An observation that cannot change the decision is worthless however informative it seems. Cost, probability and state counts are inputs to the calculation rather than the value. Asking whether an answer would change the decision, before paying to obtain it, is a principle that applies well beyond planning.

Your Answer : (Not Answered)

Correct Answer : A

Adding a symmetry breaking constraint, or using a system such as cutting planes that has short proofs where resolution does not, changes the problem the solver faces. Waiting is hopeless against an exponential, other methods may succeed, and shrinking the instance avoids the question. Changing the formulation when the method is provably inadequate, rather than persisting with it, is the same lesson that reformulation teaches in ordinary search.

Your Answer : (Not Answered)

Correct Answer : B

Since a conflict driven solver produces a resolution proof, lower bounds on the length of resolution proofs for a family of formulas bound how quickly any such solver can decide them. Certain pigeonhole formulas require exponentially long resolution proofs and defeat every such solver accordingly. Replacement, first order use and polynomial termination do not follow. Theory bounding what engineering can achieve, rather than merely describing it, is an unusually useful kind of result.

Your Answer : (Not Answered)

Correct Answer : A

Algorithms such as WalkSAT start from a random complete assignment and repeatedly flip a variable from an unsatisfied clause, mixing greedy and random choices. They solve very large satisfiable instances quickly and can never prove unsatisfiability. The other options describe unrelated procedures. The same division between complete systematic methods and incomplete local ones, with the same asymmetry about proving absence, appears here as in every other search setting.

Your Answer : (Not Answered)

Correct Answer : C

Domain constraints hold in every situation and therefore constrain the effects of every action, which is how indirect consequences are derived without enumerating them. Permissions, value ranges and scope boundaries are different notions, though value ranges are a special case. Distinguishing a fact that happens to hold from one that must hold in every situation is what allows the second to do inferential work the first cannot.

Your Answer : (Not Answered)

Correct Answer : D

A knowledge graph stores entities as nodes and typed relations as edges, at a scale of billions of assertions, and is used to answer queries and to ground search results. It is the descendant of the semantic network, given a formal basis and industrial scale. Growth plots, rule diagrams and proof structures are different objects. That an idea criticised for lacking semantics returned at enormous scale once the semantics were supplied is a fair summary of the field's history here.

Your Answer : (Not Answered)

Correct Answer : C

A player who cannot see the opponent's cards must reason about every hand consistent with the play so far, so the effective state is a distribution and the problem grows enormously. Sampling the possible states and solving each is a common approximation. Branching, finiteness and terminal counts are unaffected or worsened. Lifting a problem from states to distributions over states is the same move partial observability forces in ordinary search, and it costs the same exponential.

Your Answer : (Not Answered)

Correct Answer : B

When a node is abandoned because it cannot affect the root, its exact value was never determined and all that is known is that it lies on one side of the window. Storing it as though it were exact corrupts later searches. Evaluation accuracy, transpositions and depth are different issues. Recording what is actually known, rather than what was returned, is the discipline that keeps a caching scheme sound.

Your Answer : (Not Answered)

Correct Answer : D

Because states are described by variables, an algorithm can detect that a partial assignment is already inconsistent and prune every completion of it at once, which an atomic representation cannot express. Heuristics such as most constrained variable are a further benefit of the same structure. The space is the same, and exact solutions are returned. The lesson that representation determines what can be exploited, established early in the subject, pays off concretely here.

Your Answer : (Not Answered)

Correct Answer : C

Given the nodes generated and the solution depth, the effective branching factor is the value that would produce that many nodes in a uniform tree of that depth, and it is the standard way of comparing heuristics. A value near one indicates an excellent heuristic. The average legal action count is the raw branching factor, and the other options describe extremes and graph search effects. Summarising a search's efficiency in a single interpretable number is what makes the measure useful.

Your Answer : (Not Answered)

Correct Answer : B

Working backwards needs the predecessors of a state, which is easy in a symmetric puzzle and awkward or impossible when actions are not invertible or when the goal is described by a property rather than a specific state. Interleaving on one processor is routine, meeting is detected by intersecting the frontiers, and memory is dominated by the smaller exponents. A technique whose applicability depends on a structural property of the problem must be checked against that property first.

Your Answer : (Not Answered)

Correct Answer : A

A problem is fully specified by where the agent starts, what it can do, what each action leads to, how to recognise success and what a path costs. Everything else, including the frontier, the search tree and the choice of algorithm, belongs to the solution rather than the problem. Heuristics belong to informed search. Separating the problem from the method is what allows the same problem to be attacked by several algorithms and compared fairly. A single measurement chosen to respond to the one asymmetry beats several that respond to shared behaviour.

Your Answer : (Not Answered)

Correct Answer : D

Rewarding an outcome states what is wanted and lets the agent discover how to achieve it, which is the whole advantage of a declarative specification. Prescribing actions freezes the designer's assumptions into the system. Measurability often runs the other way, actions are usually observable, and enumeration is a practical rather than a principled objection. Saying what rather than how is a principle borrowed from declarative programming and it applies here with the same force.

Your Answer : (Not Answered)

Correct Answer : C

By requiring the performance measure to be written down first, and written in terms of outcomes rather than behaviour, the framework confronts the designer with the objective before any code exists. That is where the most expensive errors are made and the hardest to detect afterwards. Algorithm efficiency, language choice and data volume are recoverable mistakes. An error in the specification is reproduced faithfully by every subsequent stage, which is why catching it first matters so much.

Your Answer : (Not Answered)

Correct Answer : C

Updating internal state needs a transition model saying how the world changes on its own and how it changes in response to the agent's actions. Together they let the agent maintain a picture of what it cannot currently see. Sensor and actuator models refine the same idea, goals and utilities belong to higher architectures, and other agents are a special case of world evolution. Naming exactly what knowledge an architecture requires shows immediately how much work building one will be.

Your Answer : (Not Answered)

Correct Answer : A

Chess without a clock presents the whole board to both players, has fixed rules, waits for the player to move and has finitely many positions, which is why it was studied so early and so successfully. Driving fails every criterion, poker is partially observable and stochastic, and a vacuum cleaner is partially observable and continuous. Classifying a familiar task against the dimensions is the quickest way to fix them in mind.

Your Answer : (Not Answered)

Correct Answer : A

The empirical finding that performance improves predictably with data and compute has driven the scale of modern systems and shifted the field's centre of gravity towards learning. It does not remove the need for architectural and algorithmic advances, which have continued in parallel. Storage cost and interpretability are separate matters, though interpretability generally worsens with scale. A reliable empirical relationship between resources and performance is unusual and it has shaped a decade of research strategy.

Your Answer : (Not Answered)

Correct Answer : C

The agent keeps an internal state updated by a model of how the world evolves and how its own actions affect it, which lets it act sensibly when the percept alone is insufficient. Goals belong to the next architecture up, storing every percept is neither necessary nor practical, and action distributions belong to policies. Adding just enough memory to overcome partial observability, rather than storing everything, is the design idea.

Your Answer : (Not Answered)

Correct Answer : A

Critics observe that a system might pass by evasion, mimicry of human error and conversational trickery without any of the capabilities the test was meant to detect, and that the test is anthropocentric and unhelpful as a research goal. Judges are readily found, and arithmetic speed and memory are not what it measures. A test that can be gamed measures the gaming as much as the capability, which is a difficulty every benchmark in the field has faced since.