Artificial Intelligence: PSYCH homework

Artificial Intelligence: PSYCH homework

User_7975 added on 02/12/16 at 12:43 AM (PST):

1.
1. Instructor presentation:
1. Why isnt normal computing considered intelligent?
2. What is intelligence?
3. Can intelligence be demonstrated without knowledge?
4. Can animals demonstrate intelligence?

2. Buchanan (2005): A Brief History of AI
No review questions

3. Warwick (2012): Artificial Intelligence: the Basics (pp. 1-174)
Chap. 1 Review Questions
Instructor comment: recent progress in AI has been largely due to increased computer processing speed and better constraints placed on the problem to be solved.
Instructor comment: IQ tests are not subjective, as the author claims. There is much research correlating IQ with other measures of intelligence. Also, genetics is a far greater contributor to IQ than environment.
5. How can instinctual behavior be considered intelligent?

Chap. 2 Review Questions
Instructor comment: the separation of data from the program in expert systems was an important early development in AI. Expert system shells allow new systems to be developed simply by adding a new set of rules.
Instructor comment: experts usually agree on rules for problem solving. It is not as subjective as the author implies.
6. What is classical AI?
7. What does the author mean by saying that classical AI is top-down?
8. What are fuzzy rules good for?
9. What is search and how does it relate to problem solving?
10. What is knowledge representation for? What types are common?

Chap. 3 Review Questions
11. How does weak AI differ from strong AI?
12. What is Searles Chinese room argument?
13. What is the Turing Test?

Chap. 4 Review Questions
14. Why are neural nets considered a bottom-up approach to AI?
15. What does a genetic algorithm represent?
16. What is the purpose of AI agents?

Chap. 5 Review Questions
Instructor comment: cellular automata are meant to demonstrate how simple mechanisms can result in complex change, a la Darwinism. I think the whole topic of A-life is silly and useless.
17. What do AI researchers hope to gain from artificial life models?
18. How does swarm intelligence work?

Chap. 6 Review Questions
Instructor comment: Warwick seems to think that artificial intelligence and artificial life have much more to offer than biological life.
19. What do computer vision systems look for in digital images?
20. Why do mobile robots use a variety of sensors to find their way around?

2.
1. Instructor presentation:
1. How do AI researchers and cognitive scientists differ in their approach to artificial intelligence?
2. If the brain can store more information than the largest computer system, what does that say about our current technology in A.I.?
3. Why is the associativity of human memory so valuable? Does computer memory have it?

2. Chatham (2007): 10 Important Differences Between Brains and Computers
No review questions

3. Thagard (2005): Mind: Introduction to Cognitive Science (pp. 1-143)
Instructor comment: this book is especially valuable for its discussion of different types of memory representations. In order to make knowledge usable for artificial intelligence, it has to be structured in a form that preserves its facts and relations while rendering it tractable for machine computation.

Chap. 1 Review Questions
1. How does cognitive science differ from AI? How does it differ from experimental psychology?
2. What is the Computational-Representational Understanding of Mind (CRUM)?

Chap. 2 Review Questions
3. What is abduction?
4. Where does the meaning of the symbols used in logical formalisms come from? How are they represented in logic knowledge representations?
5. What do you know that is hard to express in formal logic?

Chap. 3 Review Questions
6. What is a heuristic and how is it used in problem space search?
7. How is a rule like a stimulus-response pairing?
8. Would a problem solving machine composed of a data base of rules and a means for effectively choosing which rules to activate and when constitute an intelligent system?

Chap. 4 Review Questions
9. Can a concept be completely defined by a frame memory representation with its slots and values?
10. What is the value of inheritance in hierarchical frame representations?
11. Can frames represent specific things or only generalized classes of things?

Chap. 5 Review Questions
12. How does our understanding of how water flows through pipes relate to our conception of how electricity flows through a circuit? What if we thought that blood in our circulatory system was like oil in an auto engine? (We did at one time.)
13. How could the correspondence between a sea slug and a professor be represented in a case-based reasoning knowledge representation?

Chap. 6 Review Questions
14. Is the analogy between the tumor problem and the fortress capture strategy easily representable with a verbal knowledge representation?
15. Could inventors devise new mechanisms mentally using verbal knowledge representations? Is there a need for a non-verbal knowledge representation?

Chap. 7 Review Questions
Instructor comment: verbal knowledge representations are referred to as symbol systems. Neural net representations are called sub-symbolic systems since multiple nodes, links, and weights are used to represent one output.
16. What is parallel constraint satisfaction?
17. What happens when a neural net is trained?
18. Why does backpropagation learning require a supervisor?

Chap. 8 Review Questions
Instructor comment: why should the author of Mind expect that any theory of mind could be adequate to explain reasoning when so little is known about the brain?
19. Is symbol manipulation the same thing as thinking? How is the meaning of a symbol represented?

4. Yam (1998): Intelligence Considered
Instructor comment: IQ is the psychometric (test-taking) measurement of intelligence. Many outside the field are critical of the notion of general intelligence (g) and of the finding that different racial groups have significantly different scores on IQ tests. The critics are motivated by political ideology the science of IQ is well-founded.
20. What would be the value of an AI system with generalized intelligence over a system with only specialized intelligence?

3.
1. Instructor presentation
1. What is the Physical Symbol System Hypothesis?
2. Do you think that everything you know could be represented as a symbol? Why or why not?
3. What types of knowledge representations work well with knowledge about objects?
4. What types of knowledge representations work well with procedural knowledge?
5. Why dont semantic nets work well for event knowledge?
6. What is the frame problem?
7. Why cant any of these knowledge representation methods solve the frame problem?

2. Cawsey, Alison (1998), The Essence of Artificial Intelligence Chaps. 1 & 2 (pp. 1-38)

Chap. 1 Review Questions
8. What do you think of the observation that present AI systems using
mathematical processes are more successful performers than systems based on classical knowledge modeling?

Chap. 2 Review Questions
9. Why are default slot values useful to frames when a slot value is undeclared?
10. Why is the mechanism of inheritance useful for frames and semantic nets?
11. What is the value of being able to derive facts from a knowledge base by inference over that of programming a rule or statement to cover every circumstance?
12. What advantage does predicate logic have over propositional logic?
13. What can predicate logic do that frames cannot?
14. What does predicate logic have trouble representing?
15. What kind of knowledge do frames and semantic nets represent best?
16. What is the primary difficulty with predicate logic systems?
17. What is procedural or performance knowledge and why are rules adept at representing it?
18. Why is a formalized knowledge representation method needed? Why not use natural language statements for a knowledge base?

3. Mller (2007): Is There a Future for AI Without Representation?
19. What is layered subsumption architecture?
20. Is Brooks right to claim that intelligence can be shown by a reactive system through its interaction with the environment without using knowledge representations?
21. What does Brooks mean by saying that all symbolic knowledge representations lack grounding. What is symbol grounding?
22. If a central executive function using representations is unnecessary for intelligent action, how could the autonomous vehicles in the DARPA challenge have found their way to the goal?

4.
1. Instructor presentation

1. How can problem-solving be defined as search through a problem space?

2. What is a search tree.

3. What is combinatorial explosion and how does heuristic search help deal with it?

4. Is chess-playing a well-structured or an ill-structured problem? Why?

2. Cawsey, Alison (1998), The Essence of Artificial Intelligence Chap. 4 (pp. 68-95)

Chap. 4 Review Questions
5. How does breadth-first search work? How does depth-first search work?
6. When is breadth-first search advantageous? When is depth-first search advantageous?
7. What is heuristic search? Is it exhaustive?
8. How could an evaluation function know that one node is closer to the solution and assign it a better score?
9. How is the cost function in A* search different from the evaluation function in Best First search?
10. In the water jug problem, the representation of the problem and the algorithms used to solve it are programmed. The program apparently runs from start to finish in the same way every time. Does the water jug program demonstrate artificial intelligence or is it no different than any algorithmic computer program?
11. What kind of search is employed in the water jug problem?
12. The water jug program demonstrates no understanding of the process of devising a solution. Does it solve the problem by trial and error? Would this approach have limitations when applied to other problems?
13. How does planning make a complex problem more tractable?
14. How does solving the preconditions in the robot-fetch beer problem resemble the backwards chaining of rules?
15. In the robot-fetch beer problem, how does the system know that the door must be open before the robot can enter the kitchen?

2. Schaeffer (2001), A Gamut of Games
16. Why have game-playing programs had such success?
17. What is the alpha-beta search algorithm that Schaeffer credits for the success of a number of game-playing programs?
18. Why is brute force search, which used to be avoided, now often a viable option in game-playing programs?
19. How is the metric of cost measured in some game-playing programs and how is it used?
20. How are Monte Carlo simulation and temporal-difference learning alike?
21. Do game-playing programs which model human game performance truly demonstrate intelligence?
22. How has increased computer processing speed enhanced the performance of game-playing programs?
23. Is the chess knowledge incorporated in Deep Blue useful for any other purpose?
24. What is the primary reason why the best game-playing programs now outperform the best human players?
25. Why have the game of Go playing programs been less successful than other game-playing programs

3. Klein (1996): Nonlinear Aspects of Problem Solving
26. What are ill-structured problems?
27. Why doesnt the model of search through a problem space work for ill-structured problems?
28. According to Klein, ill-structured problems require continual redefinition of the problem and constructing options from leverage points. Do you know how this could be implemented in a knowledge representation?

4. Halfhill (1997): Searching for Deep Blue
29. What does the fact that chess grandmasters may consider only three moves per second while Deep Blue can consider 100 million moves per second tell us about our understanding of human problem solving in the game of chess?
30. Do you think that the technology of Deep Blue would be useful in solving other problems? What kind of problems?
31. Do you think that increased computer speed alone will solve most of the present shortcomings of AI?
5.
1. Instructor presentation
1. Why are expert systems considered strong methods of problem solving?
2. Why can expert systems often perform as well as human experts?
3. Why is it important to limit the knowledge domain of an expert system?
4. Why can the same inference engine be used in different expert systems?
5. How does backward chaining work?
6. What is a certainty factor?

2. Cawsey, Alison (1998): The Essence of Artificial Intelligence Chap. 3 (pp. 40-65)
Chap. 3 Review Questions
7. What does a knowledge engineer do? How is knowledge acquired from experts?
8. Why are rules the preferred method of knowledge representation for expert systems?
9. Why is the capability for reviewing how the expert system reaches conclusions important? Why are rules ideal for this purpose?
10. What kinds of problems should expert systems be used for? What kinds of problems should they probably not be used for?
11. What kind of problem would be best for a forward-chaining system? What kind of problem would be best for a backward-chaining one?
12. Why are rule bases often considered shallow knowledge representations?
13. What does Bayes Theorem allow us to do?
14. What is the questionable assumption that must be made if conditional probabilities are used? How does it affect the conclusion arrived at?
15. How are certainty factors different from probabilities? How are certainty factors determined?
16. Why might the Internist system be said to use bidirectional chaining of rules?
17. Why would the belief networks used in Pathfinder be considered deeper knowledge representations than rule bases?
18. Why do you suppose that expert system development was expensive in the early years?
19. Why do you suppose that medical expert systems are not widely used for diagnosis?
20. How would a general problem solver try to solve medical diagnosis problems with means-end analysis?

3. Talebzadeh, et al. (1995): Countrywide Loan Underwriting Expert System
Instructor comment: Countrywide was caught up in the sub-prime loan scandal of 2008 and was convicted of making special loans to Democrat politicians. CLUES apparently approved 100% of government insured loans.
21. Why were case-based reasoning and neural nets rejected for the knowledge representation role for this expert system?
22. What difficulties occurred for the knowledge engineers because the underwriters used heuristics instead of rules to make their loan decisions? How did they acquire knowledge from the experts?
23. What would the knowledge engineers have done if the loan experts had disagreed?
24. A universal problem for expert systems is user confidence and acceptance. How did the CLUES developers overcome this problem?
25. In the validation phase of development, why is it important that the reasoning used by the expert system be transparent and that it match the reasoning used by the experts?
26. Could expert systems potentially do for bureaucratic jobs what robotics has done for manufacturing?
Instructor comment: the CLUES developers could have validated the rules used for loan decisions with regression analysis and saved some time. Working as closely as they did with the underwriters, however, must have improved user acceptance of the system.
6.
1. Instructor presentation
1. Why dont word for word translations work when translating a language?
2. How does understanding context help us interpret text? Do you think that NLP uses context?
3. Do you think that natural language programs can understand metaphors and analogies? Does NLU understand the meaning of language at all?
4. Why arent chatterbots true natural language processing programs?
5. How did SHRDLU differ from chatterbot programs?
6. What does a script do in an NLP program?
7. How does a classic story understanding program understand a news story?
2. Cawsey, Alison (1998): The Essence of Artificial Intelligence Chap. 5 (pp. 98-122)
Chap. 5 Review Questions
Instructor comment: this author includes speech recognition with natural language processing. They have been studied separately in the history of AI, although speech understanding is really much the same problem as natural language understanding.
8. What problem does speech recognition have that natural language processing doesnt?
9. How is common sense knowledge often necessary to disambiguate sentences? Why is this a problem for AI to accomplish?
10. How does syntax help establish the meaning of a sentence for natural language processing?
11. How does a language parser know which words in a sentence are nouns, verbs, or determiners?
12. In the semantics analysis phase, how are the meanings of words that are added in this stage of processing determined? (I cant find out how from reading this section.)
Instructor comment: the author does not fully address how word meaning is derived through information in the knowledge base.
13. How could the meaning of love be represented in a knowledge base?
14. What are pragmatics in NLP?
15. What kind of knowledge is necessary to determine correct pronoun reference? Can parsing provide it?
3. Feldman (1999): NLP Meets the Jabberwocky
16. What are the different aspects of language that help to define the meaning of sentences?
17. Why can humans interpret ambiguous language without difficulty while natural language understanding systems find it difficult?
18. Why cant information retrieval systems retrieve documents based on their meaning rather than using citation ranks, word frequencies, and correlations?
19. An NLU system performs semantic analysis by looking up associated words. Is this all that is needed for an understanding of semantics?
20. The author says that an NLP information retrieval system can use context in the query to disambiguate search terms and eliminate irrelevant search items. She doesnt say how. Do you know how the context of the query could be found?
Instructor comment: this article was published in 1999, before Google became so successful. I dont think that Google uses Natural Language Understanding in its search algorithms. It is a statistical beast. My nephew is a Ph.D. in statistics and works for Google.
4. Lenat (1995): A Large-Scale Investment in Knowledge Infrastructure
21. Why did Lenat choose to enter specific axioms into his knowledge base rather than general axioms?
Instructor comment: CYCs knowledge base used to consist of millions of rules. They were changed to a kind of predicate calculus to make them more efficient, I believe, and perhaps to allow for inferencing.
22. Why doesnt CYC use certainty factors in its axioms?
23. Why is every axiom in CYC tied to a particular context?
24. Do any of the applications Lenat envisions for CYC seem to be solvable by other kinds of software solutions?
25. How could CYCs knowledge be used to make characters in a role-playing game more lifelike? How could it make them more spontaneous, unpredictable, adaptable, and crafty?
Instructor comment: I could find no more recent article of a general nature about CYC than this one from 1995. Lenat has seemed to be less than forthcoming about the progress or lack of it being made by CYC.
5. Ferrucci, et al. (2010): Building Watson an Overview
26. Does Watson sound like it could offer an alternative in Natural Language Processing to CYC?
27. Why should IBM undertake projects such as Deep Blue if the technology is not transferable to other applications?
28. What does the Natural Language Understanding component of Watson do?
29. Is Watson just a super search engine? How does it demonstrate intelligence?
30. How would Watson know that two words rhyme?
31. Watsons IBM predecessors did not work especially well. What was the biggest contributor to its breakthrough in performance?
32. Do you think that massive parallelism, many independent experts, and confidence estimation will be required for success in question answering systems from now on?
Instructor comment: many existing text processing technologies were combined to make Watson. The overall approach used was to employ all of them to generate multiple candidates for answers and then score the candidates with evaluation functions. These text processors often looked for patterns in a corpus of human sentences, or, as in the case of the final scoring function, a collection of Jeopardy questions and answers. I wonder if this is how artificial intelligence will get its smarts in the future not from programming it in, but through extracting it from humans through machine learning.
33. Do you think that the question classification system within Watson contains specialized heuristics for answering each kind of question in Jeopardy? Will heuristics and programming tricks always be needed in any AI system?
34. Is there any real understanding of semantics anywhere in Watson?
35. Could Watson be said to use a swarm approach to question answering, i.e., a great number of experts generating results and communicating them to other experts in the swarm?
36. How do you think that Watson detects relations between objects in a database?
37. The question decomposition function that relies on computed confidence scores is conceptually similar to heuristic search in problem solving. How?
38. Watson generates hundreds of hypothetical answers for each Jeopardy question. How does it decide which one is right?
39. Can Watson be considered a Natural Language Understanding program? Why or why not?
40. Do any of the scoring algorithms used in Watsons hypothesis scoring have any understanding of the meaning of the candidate answers?
41. If it takes a system like Watson with massively parallel processors, hundreds of algorithmic routines, and a large team of developers striving for years to make it a success at one specialized task, wont these systems become too expensive for general use?
42. How was machine learning used to improve Watsons final answer scoring function?
43. Can you think of other areas of AI that may benefit from applying the Watson approach to AI: massively parallel processing employing a vast array of probabilistic algorithms tuned to produce the desired output from a particular class of inputs?
44. With reference to Searles Chinese Room argument, can Watson be said to understand the Jeopardy questions it answers?
7.
1. Instructor presentation:
1. What is speech synthesis?
2. Which parts of the vocal apparatus produce consonant sounds?
3. What is a formant?
4. Are phonemes the same as vowels or consonants? If not, what are they?
5. What is phoneme co-articulation and why is it a problem for speech recognition?
6. What is surprising about the Phonemic Restoration Effect?
7. Why is discrete speech easiest to recognize?
8. What does statistical modeling do to improve speech recognition performance?
9. What does speech understanding try to do? Is it much different than natural language understanding?
2. Cawsey (1998): The Essence of Artificial Intelligence Chap. 5 (pp. 98-103)
Instructor comment: this chapter was already covered in Section 6: Natural Language Understanding.
3. Deng and Huang (2004): Challenges in Adopting Speech Recognition
10. What do the authors attribute progress in speech recognition to over the past 30 years?
11. According to the authors, what has to be done to increase the acceptance of speech recognition systems?
12. Why dont speech recognition systems work well in noisy environments? Why is this less of a problem for humans?
13. How will adding a true understanding of context and common sense improve speech recognition?
14. Why is conversational speech difficult for speech recognition systems to process accurately?
4. Furman, et al. (1999): Speech-Based Services
15. Why is accuracy in word recognition such a limiting factor in the user acceptance of speech recognition systems?
16. Errors in the recognition of speech will occur. How does the Lucy system try to minimize the inconvenience?
17. Why cant word templates used to identify single words be expanded to recognize continuous speech?
18. Do continuous speech recognition algorithms identify individual words?
19. What is wordspotting and what does it enable a speech recognition system to do with connected speech?
20. How does grammar-based speech recognition reduce errors in processing connected speech?
21. What is a finite-state grammar map? Why is it tedious to construct and not generalizable across applications?
22. How does statistical language modeling work? What advantage does it have over finite-state grammars?
23. What limitation does statistical language modeling have on the range of utterances it can handle?
24. According to the authors, what will have to be accomplished before unconstrained speech understanding is attained?

5. White (1990): Natural Language Understanding and Speech Recognition
Instructor comment: this article, like many technical articles, does not explain the technical jargon used. Therefore, parts of the text are dense and hard to understand for those of us unfamiliar with the research.
25. Why havent speech recognition systems simply been merged with Natural Language Understanding systems to create speech understanding?
26. Why is connected speech so difficult for speech recognition?
27. Why do you think humans can recognize connected speech effortlessly, while speech recognition systems have such difficulty?
28. What are
29. What is prosody? Can it be incorporated into speech recognition?
30. Does this article adequately define context and how it constrains the interpretation of speech? Why or why not?
31. What problem does ungrammatical speech pose for speech recognition?
32. How does the author envision using higher level speech knowledge (semantics, pragmatics, prosodics) to enhance system performance at the phoneme level?
33. Why does the author think that the ability to combine and coordinate higher level knowledge sources through NLU merging with ASR will lead to progress in speech recognition?

8.
1. Instructor presentation
1. Why did early A.I. researchers think that computer vision would be easy to solve?
2. Why have industrial vision systems been successful at vision tasks for many years now?
3. How does our understanding of the world help us recognize objects?
4. Is the human eye more complex than a video camera? How does the eye pre-process images before sending them to the brain?
5. What is top-down image processing and how does it differ from the bottom-up processing of computer vision?
6. How do we know that the brain probably does not store images as photographic templates?
7. Why does computer vision require tremendous processing power?
8. Do you think that face recognition technology could be used to identify all objects?
9. Do you think that deep learning statistical methods will finally solve the problems of computer vision?

2. Cawsey (1998): The Essence of Artificial Intelligence Chap. 6 (pp. 125-141)
Chap. 6 Review Questions
10. Do vision systems necessarily have to identify objects to provide useful information?
11. Smoothing of the pixellated image contours blurs the outline and features of the image. Why is this necessary in present systems? What might the implications of his practice be for the development of improved object recognition?
12. What do masking arrays and difference operators do with the pixel intensity values on the digitized image?
13. Finding edges and line segments is computationally intensive yet yields poor results. Would it be wise to stay with this approach and wait for more powerful computers or devise some other way of analyzing visual scenes?
14. Objects in a scene often occlude each other. Would the image analysis routines described need some additional information to clarify the jumble of contours in a scene?
15. Does the use of texture and shading to determine the orientation of an object depend first on identifying which surfaces are connected together in the same object? What would happen to this information if the edge contours of the object could not be determined?
16. What kind of knowledge representation could act to identify an object based on the information provided from an image analysis? How could that represen-tation be general enough to recognize varying types and sizes of the objects that it defines (e.g., houses, dogs, and trees)?
17. Contextual information has been used in A.I. to constrain the number of possible interpretations of the data. How could it be applied to computer vision?
18. How is the practice of using models for object recognition like using templates for speech recognition?
19. What level of detail in a visual model would be necessary to distinguish a dog from a cat? Would this defeat the purpose of using a simplified, computationally tractable model?
20. How do industrial vision systems constrain the complexity of visual tasks? Why is industrial vision not generally considered A.I., but rather, an engineering domain?

3. Piccardi and Jan (2003): Recent Advances in Computer Vision
21. Will human-computer interfaces that use video sensing to respond to hand gestures and speech recognition to respond to audio commands necessarily replace the keyboard and mouse?
Instructor comment: digital cameras now find faces in scenes and draw boxes around them high tech for only 10 years ago!
22. How are people identified in a surveillance video? Does this provide sufficient information to tell one person from another?
Instructor comment: a neural net was used in the surveillance system mentioned in the article to distinguish one pattern of target movement from another. We will discuss how this is done in Section 9 on Neural Nets.

4. Pinz (2005): Foundations and Trends in Computer Graphics and Vision
23. Why is object recognition the primary goal of A.I. vision research?
24. What is the difference between generic object recognition and specific object recognition?
25. Humans effortlessly identify objects and the general category they belong to. Why is it so hard for computer vision?
26. If humans are so adept at object recognition, why are industrial vision systems faster and more accurate at inspection tasks? Is inspection a kind of object recognition?
27. Since computational intractability is a bugaboo of vision understanding, why would some researchers choose to compound their problems by working with 3D instead of 2D visual representations?
28. How can context help determine the identity of a visual object?
29. Specific object recognition (e.g., identifying people from facial images) appears to consist of matching designated features and measures between features on the image to features and measurements on a stored reference model. Why will this approach not work with generic object recognition?
Instructor comment: the appearance-based approach to image recognition for faces mentioned in the article compares pixel values and locations on a bit-mapped image to a template statistically derived from a series of training images.

5. Torralba, et al. (2010): Using the Forest to See the Trees
Instructor comment: using the context of visual scenes to constrain the recognition problem seems a no-brainer, but it was apparently not investigated until recently. Also, directing attention to areas of interest in the visual scene should have been a focus of research from the start instead of relying on brute force to process every pixel.
30. What is the problem with requiring one specialized detector for every object you want to identify?
31. What does the gist descriptor do to create a representation of a scene that does not require the identification of its components?
32. The detection of the presence of an object and its location in a scene were apparently state of the art accomplishments in 2010. What kind of vision application is ideal for that technology?
33. How is object localization performed by the method described by the authors?
Instructor comment: understanding the math in this article isnt necessary for our purposes. We want to see what this system can do and what the technology is like behind it.
34. What did the authors combine to make their integrated vision system?
35. What did the result of processing the third image in Figure 6 (6b at the bottom) show about the detectors and gist descriptor employed?
Instructor comment: the nature of the oracle that labels scene categories and enables the best performance is not explained.
36. How does knowledge of the scene category help determine the presence of cars in it?
37. Occlusion of objects in images can make them very difficult to identify, as the results discussed show. What information do humans make use of to identify occluded objects that computer vision systems dont?
38. What method do the authors use to try to reduce false positives, i.e., finding cars where none exist?
39. Do you believe that mathematical models will eventually extract enough information from a scene to make accurate classifications of scenes and recognition of objects, or must something else be added before that can happen?



User_7975 added on 02/12/16 at 12:45 AM (PST):

Are you comfortable with artificial intelligence?? Let me know if you have any questions!!
Thanks!!

User_7975 added on 02/12/16 at 02:06 AM (PST):

This is one piece/chapter

1. Instructor presentation:
1. Why isnt normal computing considered intelligent?
2. What is intelligence?
3. Can intelligence be demonstrated without knowledge?
4. Can animals demonstrate intelligence?

2. Buchanan (2005): A Brief History of AI
No review questions

3. Warwick (2012): Artificial Intelligence: the Basics (pp. 1-174)
Chap. 1 Review Questions
Instructor comment: recent progress in AI has been largely due to increased computer processing speed and better constraints placed on the problem to be solved.
Instructor comment: IQ tests are not subjective, as the author claims. There is much research correlating IQ with other measures of intelligence. Also, genetics is a far greater contributor to IQ than environment.
5. How can instinctual behavior be considered intelligent?

Chap. 2 Review Questions
Instructor comment: the separation of data from the program in expert systems was an important early development in AI. Expert system shells allow new systems to be developed simply by adding a new set of rules.
Instructor comment: experts usually agree on rules for problem solving. It is not as subjective as the author implies.
6. What is classical AI?
7. What does the author mean by saying that classical AI is top-down?
8. What are fuzzy rules good for?
9. What is search and how does it relate to problem solving?
10. What is knowledge representation for? What types are common?

Chap. 3 Review Questions
11. How does weak AI differ from strong AI?
12. What is Searles Chinese room argument?
13. What is the Turing Test?

Chap. 4 Review Questions
14. Why are neural nets considered a bottom-up approach to AI?
15. What does a genetic algorithm represent?
16. What is the purpose of AI agents?

Chap. 5 Review Questions
Instructor comment: cellular automata are meant to demonstrate how simple mechanisms can result in complex change, a la Darwinism. I think the whole topic of A-life is silly and useless.
17. What do AI researchers hope to gain from artificial life models?
18. How does swarm intelligence work?

Chap. 6 Review Questions
Instructor comment: Warwick seems to think that artificial intelligence and artificial life have much more to offer than biological life.
19. What do computer vision systems look for in digital images?
20. Why do mobile robots use a variety of sensors to find their way around?


User_7975 added on 02/12/16 at 02:08 AM (PST):

Here's the 2nd piece
1. Instructor presentation:
1. How do AI researchers and cognitive scientists differ in their approach to artificial intelligence?
2. If the brain can store more information than the largest computer system, what does that say about our current technology in A.I.?
3. Why is the associativity of human memory so valuable? Does computer memory have it?

2. Chatham (2007): 10 Important Differences Between Brains and Computers
No review questions

3. Thagard (2005): Mind: Introduction to Cognitive Science (pp. 1-143)
Instructor comment: this book is especially valuable for its discussion of different types of memory representations. In order to make knowledge usable for artificial intelligence, it has to be structured in a form that preserves its facts and relations while rendering it tractable for machine computation.

Chap. 1 Review Questions
1. How does cognitive science differ from AI? How does it differ from experimental psychology?
2. What is the Computational-Representational Understanding of Mind (CRUM)?

Chap. 2 Review Questions
3. What is abduction?
4. Where does the meaning of the symbols used in logical formalisms come from? How are they represented in logic knowledge representations?
5. What do you know that is hard to express in formal logic?

Chap. 3 Review Questions
6. What is a heuristic and how is it used in problem space search?
7. How is a rule like a stimulus-response pairing?
8. Would a problem solving machine composed of a data base of rules and a means for effectively choosing which rules to activate and when constitute an intelligent system?

Chap. 4 Review Questions
9. Can a concept be completely defined by a frame memory representation with its slots and values?
10. What is the value of inheritance in hierarchical frame representations?
11. Can frames represent specific things or only generalized classes of things?

Chap. 5 Review Questions
12. How does our understanding of how water flows through pipes relate to our conception of how electricity flows through a circuit? What if we thought that blood in our circulatory system was like oil in an auto engine? (We did at one time.)
13. How could the correspondence between a sea slug and a professor be represented in a case-based reasoning knowledge representation?

Chap. 6 Review Questions
14. Is the analogy between the tumor problem and the fortress capture strategy easily representable with a verbal knowledge representation?
15. Could inventors devise new mechanisms mentally using verbal knowledge representations? Is there a need for a non-verbal knowledge representation?

Chap. 7 Review Questions
Instructor comment: verbal knowledge representations are referred to as symbol systems. Neural net representations are called sub-symbolic systems since multiple nodes, links, and weights are used to represent one output.
16. What is parallel constraint satisfaction?
17. What happens when a neural net is trained?
18. Why does backpropagation learning require a supervisor?

Chap. 8 Review Questions
Instructor comment: why should the author of Mind expect that any theory of mind could be adequate to explain reasoning when so little is known about the brain?
19. Is symbol manipulation the same thing as thinking? How is the meaning of a symbol represented?

4. Yam (1998): Intelligence Considered
Instructor comment: IQ is the psychometric (test-taking) measurement of intelligence. Many outside the field are critical of the notion of general intelligence (g) and of the finding that different racial groups have significantly different scores on IQ tests. The critics are motivated by political ideology the science of IQ is well-founded.
20. What would be the value of an AI system with generalized intelligence over a system with only specialized intelligence?

User_7975 added on 02/12/16 at 02:09 AM (PST):

And the third piece

1. Instructor presentation
1. What is the Physical Symbol System Hypothesis?
2. Do you think that everything you know could be represented as a symbol? Why or why not?
3. What types of knowledge representations work well with knowledge about objects?
4. What types of knowledge representations work well with procedural knowledge?
5. Why dont semantic nets work well for event knowledge?
6. What is the frame problem?
7. Why cant any of these knowledge representation methods solve the frame problem?

2. Cawsey, Alison (1998), The Essence of Artificial Intelligence Chaps. 1 & 2 (pp. 1-38)

Chap. 1 Review Questions
8. What do you think of the observation that present AI systems using
mathematical processes are more successful performers than systems based on classical knowledge modeling?

Chap. 2 Review Questions
9. Why are default slot values useful to frames when a slot value is undeclared?
10. Why is the mechanism of inheritance useful for frames and semantic nets?
11. What is the value of being able to derive facts from a knowledge base by inference over that of programming a rule or statement to cover every circumstance?
12. What advantage does predicate logic have over propositional logic?
13. What can predicate logic do that frames cannot?
14. What does predicate logic have trouble representing?
15. What kind of knowledge do frames and semantic nets represent best?
16. What is the primary difficulty with predicate logic systems?
17. What is procedural or performance knowledge and why are rules adept at representing it?
18. Why is a formalized knowledge representation method needed? Why not use natural language statements for a knowledge base?

3. Mller (2007): Is There a Future for AI Without Representation?
19. What is layered subsumption architecture?
20. Is Brooks right to claim that intelligence can be shown by a reactive system through its interaction with the environment without using knowledge representations?
21. What does Brooks mean by saying that all symbolic knowledge representations lack grounding. What is symbol grounding?
22. If a central executive function using representations is unnecessary for intelligent action, how could the autonomous vehicles in the DARPA challenge have found their way to the goal?


User_7975 added on 02/12/16 at 02:10 AM (PST):

I have 11 more of these for ten dollar a piece. Same page ??