サンプル問題 1: A problem-solving AI model is being developed to solve the Tower of Hanoi puzzle by systematically planning moves to achieve the final goal. The model needs to evaluate intermediate steps to clear disks strategically for future moves. Which approach would best help the model to complete this task?
A: Tactical lookahead, as it considers each immediate move without long-term planning.
B: Strategic planning, by identifying intermediate moves that contribute to the end goal.
C: Reinforcement learning, by rewarding moves that lead to progress in the puzzle.
D: Randomized search, by exploring different moves without a predefined plan.
正解:B
Explanation: Correct (B): Strategic planning is effective here as it enables the model to identify necessary intermediate steps that contribute to the overall task completion. Incorrect (A): Tactical lookahead focuses on immediate moves, but lacks the broader, goal-oriented strategy required for solving this puzzle. Incorrect (C): Reinforcement learning, while useful for trial-and-error scenarios, is less efficient for puzzles where specific sequences are needed for completion. Incorrect (D): Randomized search lacks the systematic approach necessary for achieving the goal in a structured puzzle like Tower of Hanoi
サンプル問題 2: A checkers AI program evaluates possible board positions using six criteria, each with a calculated score (c1 to c6) and corresponding weight (w1 to w6). Which formula will provide the overall score v for a given board position?
Explanation: Correct (C): v = (w1 × c1) + (w2 × c2) + (w3 × c3) + (w4 × c4) + (w5 × c5) + (w6 × c6) accurately combines the weighted values of each criterion to give an overall score for the board position. Incorrect (A): Adding weights alone does not account for the specific values of each criterion. Incorrect (B): Summing criteria values without weights ignores the relative importance of each factor. Incorrect (D): Averaging criteria values disregards individual weights, which are essential for accurate position assessment.
サンプル問題 3: Which early concept in neural network theory proposed that repeated activation of one neuron by another would strengthen their connection?
A: McCulloch-Pitts logical calculus.
B: Hebbian learning.
C: Turing computability.
D: Perceptron learning.
正解:B
Explanation: Correct (B): Hebbian learning is based on the idea that repeated activation between neurons strengthens their connection, forming the basis of associative learning in neural networks. Incorrect (A): McCulloch- Pitts focused on propositional logic to model neuron behavior, not the strengthening of neuron connections. Incorrect (C): Turing computability defines the limits of what can be computed, not neuron connection strengthening. Incorrect (D): The Perceptron was a model developed by Rosenblatt, based on probability theory rather than neuron connection strengthening.
GARP RAI試験の準備に向けた復習資料としてRAI問題集を提供しています。Risk and AI (RAI)試験の主要な知識ポイントをカバーしています。ページ上部の「RAI 試験を開始する」ボタンからオンライン無料サンプル問題を確認できます。
RAI問題集の収録内容と更新状況はどうなっていますか?
答え:現在提供されているRAI問題集には、RAI試験対策に向けた全300問の練習問題と解答が収録されています。Risk and AI (RAI)試験に向けた復習に利用できます。