Take Home Midterm Exam | Artificial Intelligence 2000 | CS 6601, Exams of Computer Science

Download Take Home Midterm Exam | Artificial Intelligence 2000 | CS 6601 and more Exams Computer Science in PDF only on Docsity! CS 6660 Intelligent Agents Fall 2000 Take-Home Mid-Term Examination Assigned: Tuesday, October 10, 2000 Due (in class): Thursday, October 19, 2000 Ashok K. Goel 1 Question#1 (100 points): Briefly explain each of the following terms: (i) Intelligent Agent (ii) Autonomy (iii) Rationality (iv) Bounded Rationality (v) Agent Architecture (vi) Reactive Control (vii) Gradient Descent (viii) Backpropagation (ix) Fitness Function (x) Blame Assignment (xi) Primal Sketch (xii) Visual Grammar (xiii) Working Memory (xiv) Productions (xv) Frame (xvi) Script (xvii) Expectation Generation (xviii) Case Grammar (xix) Spreading Activation (xx) Ontology Question#2 (50 points): 2a: What is the method of gradient descent? What (if any) is the relation between gradient descent and the delta rule for learning in perceptrons? What does this relationship tell us about the delta rule for learning? 2b: What is a linearly separable problem? Give an example of a problem that is not linearly separable. What (if any) is the relation between lin- early separable problems and the delta rule of learning? What does this relationship tell us about the perceptron method of learning? 2 Question 5 (50 points): A central dilemma in AI is that while most interesting problems are com- putationally intractable, intelligent agents have only limited computational resources. Briefly explain AI concepts and methods for addressing the issue of computational complexity. 5 Question 6 (50 points): Genetic Algorithms and Multi-Layer Neural Networks are two examples of AI methods. How may an agent decide which of these two methods to use in what situation? How might an agent more generally decide to select among AI methods available to it? 6