Due Sat Dec 18.
- Final Project Writeup.
Due Fri Dec 17.
- Final at 8 am in regular classroom.
- Final Project Code.
Due Fri Dec 10.
- In addition to project demo itself, create a Template Project for yourself and add it to the list on the Project page. Note: Your project flyer must be posted by 9 am the day of the demo! See Discussion Group for wiki password.
Due Mon Dec 6.
- Email/tweet three questions for MM that interest you based on your SoM readings. Compose the questions as if you would ask them of MM (you might). Assume he's generally familiar with all the AIMA content we've covered throughout the class.
Due Fri Dec 3.
- Read and report on SoM Chapter 20.
Due Mon Nov 29.
- Read and report on SoM Chapter 19.
Due Fri Nov 26.
Due Wed Nov 24.
Due Mon Nov 22.
- Read and report on SoM Chapter 17.1 11, Development.
Due Fri Nov 19.
- Complete your final project proposal.
Due Wed Nov 17.
- Read and report on SoM Chapter 16.1 10, Emotion.
Due Mon Nov 15.
- Read and report on SoM Chapter 15.1 15.11, Consciousness and Memory.
Due Fri Nov 12.
- Read and report on SoM Chapter 14.5 14.9, Reformulation.
Due Mon Nov 8.
- Read and report on SoM Chapter 14.1 14.4, Reformulation.
Due Fri Nov 5.
- Create a second final project proposal, this time focusing on a problem that is addressable by using adversarial search (e.g., minimax or the optimized alpha-beta).
- I am looking for a mapping from a real-world problem to an AI approach. Please don't choose a game and then describe how you'll solve it with minimax. E.g., some of the apps from the first round that are good real-world problems are: GPS with traffic considerations, path planning for airplanes, optimizing a racing bicycle, composing a melody, and finding distance to friends in a friend network.
- I realize that this is hard. In an actual R&D scenario, you would have a palette of solution approaches from which to choose when working on a problem like this. In this case, you have the solution approach and you must find the problem. Still, constraints are good for creativity so I know you will come up with good things.
- Please do some reasonable amount of formal analysis in the proposal itself. The proposal should read like you do deeply understand the problem. In the best case, the proposal presents an interesting problem, a creative solution approach that is explained clearly and looks promising, the evaluation metrics, and the impacts. The reader should think, Yes, this is clearly worth doing, and I would support this because it will make a real contribution knowledge or This will make us lots of moneyI want to invest in this project now.
- Highlight title (e.g., in bold face). You do not need to identify the title as the title; it should be obvious from its placement at the top and that it's in bold face.
- Make sure your real name is on the document itself!
- Identify sections that are specified in requirements. This is a suggestion; it's a bit in conflict with the next suggestion
- Use narrative form. It is more engaging than just answering questions in a list.
- Don't sell yourself short on the broader impacts! Even entertainment-related projects can be lucrative. Both helping the world and making money are good things.
- Use all the space (at least, 80% of it)if it says one page, use it all.
- Make sure formatted for U.S. Letter (not A4).
- No games that are trivially mapped to formal representations.
submit fredm ai-fpp2 <your file>.
- Complete questions 4 through 9 of the Berkeley reinforcement learning project. Turn in your work using
submit fredm ai-ps3b <file[s]>.
Due Wed Nov 3.
- read and comment on this article:
Due Mon Nov 1.
- Read and report on SoM Chapter 13 Seeing and Believing.
Due Fri Oct 29.
- Read and report on SoM Chapter 12.7 12.13, Learning Meaning.
- Prepare Final Project Idea Pitch #1. This is a 1-page project summary of an idea that you would consider turning into a final project. You will produce three or four of these project pitches before deciding on your final project. Just because you propose something now does not imply that you are obligated to produce it later. On the other hand, I expect that each pitch be a serious and plausible idea that you are actually interested in.
- Project title. No more than ten words.
- Problem statement. What is the problem and why is it interesting?
- Problem analysis. Explain why this problem is amenable to solution via informed search. You don't need to fully solve it (e.g., you don't need to already know exactly how you will represent it as a search problem, or what admissible heuristic you would use), but you need to argue that it's amenable to solving via search.
- Data set. What data set will you use as the basis for the analysis? Where will you get the data (e.g., you will download it from a web site, you will create it via a simulation that you build, etc.)?
- Analysis of results. How will you know if you are successful?
- Broader impacts. Why is this problem worth solving? Who will care about the solution? What impact on research or society will the work have?
submit fredm ai-fpp1 <your file>.
Due Wed Oct 27.
- Read and report on SoM Chapter 12.1 12.6, Learning Meaning.
Due Mon Oct 25.
- Please listen to this 3-minute NPR story, and write 3 "tweets" about it:
Due Fri Oct 22.
- Quiz 1 will be administered in class. The quiz will cover all course material through and including problem set 2. Society of Mind through Chapter 10 may be examined. Problem set 3 (MDPs) will not be included.
Due Wed Oct 20.
- Complete questions 1 through 3 of the Berkeley reinforcement learning project. Turn in your work using
submit fredm ai-ps3a <file[s]>.
- Read and comment on SoM Chapter 11.5 11.9, The Shape of Space.
Due Mon Oct 18.
- Read and comment on SoM Chapter 11.1 11.4, The Shape of Space.
Due Fri Oct 15.
- Read and comment on SoM Chapter 10.5 10.9, Paperts Principle. Make sure to put umlai in the Subject line!
Due Wed Oct 13.
- Finish the Berkeley multi-agent project, completing problems 4 and 5. Turn in using
submit fredm ai-ps2b <file[s]>.
Due Tue Oct 12.
- Read and comment with three "tweets" on Brains and Bytes (D. Lindley, Communications of the ACM, 53(9), September 2010, pp. 1315). Note: you must be signed into the course Google group to download. If you can't log into the Google group, download from here (either, on-campus or using your
@student.uml.eduemail/password). Note 2: Remember to put
umlaiin the Subject line if emailing.
Due Fri Oct 8.
- Complete problems 1 through 3 of the Berkeley multi-agent project. Turn in your work using
submit fredm ai-ps2a <file[s]>.
- Read the NYT article, SMARTER THAN YOU THINK: Aiming to Learn as We Do, a Machine Teaches Itself, and write three 140-char statements about its major points or your interpretations thereof. (Here is a link to a PDF.)
Due Wed Oct 6.
- Read and report on SoM Chapter 10.1 10.4, Paperts Principle.
Due Mon Oct 4.
- Read and report on SoM Chapter 9, Summaries.
Due Fri Oct 1.
- Everyone. Read and report on SoM Chapter 8.
- Grad students. Using historical notes at the end of the chapter(s) on search, research origins of at least one particular search methodology. Find an original paper where the method(s) were published. Describe the context of the problem on which the original researchers were working. Then, find a more modern paper extending recent search techniques. Also describe the situation that led to the work, and the nature of the innovation. Write up your findings in a 2- to 3-page paper (single spaced). Include full paper citations (see bibliograph of AIMA for standard bib format example).
submit fredm ai-paper1 filename.[doc,docx,pdf,ps,txt] .Please note the only acceptable file formats are as indicated in the example submit.
submitis now working on the
Due Wed Sep 29.
- PROJECT 1C. Complete questions 6, 7, and 8 in the Berkeley search project. Turn code in using
submit fredm ai-ps1cfile(s) .
Due Mon Sep 27.
- Read and report on SoM Chapter 7, Problems and Goals.
Due Wed Sep 22.
- No SoM reading due.
- PROJECT 1B. Complete the following questions in the Berkeley search project:
- Question 2, implementing breadth-first search.
- Question 3, implementing uniform-cost search, and using the
StayWestSearchAgents with their provided cost functions.
- Question 4, implementing A* search using Manhattan distance as the cost heuristic, and comparing its efficiency to UCS.
submit fredm ai-ps1b search.py searchAgents.py
Due Mon Sep 20.
- Read and report on SoM Chapter 6.
Due Fri Sep 17.
- Read and report on SoM Chapter 5.
Due Wed Sep 15.
- PROJECT 1A. Complete Question 1 in the Berkeley search project. This is the implementation of depth-first search. Note: even though this question is only worth 2 points, it represents a lot of work, and I understand that, so for us it's counting as a whole assignment.
submitcommand on Mercury, as follows:
- If you're not doing the work on
search.pyup to your account.
- Log in to
cdto the directory where the
.pyfile is located.
- Type the following to turn in your code:
submit fredm ai-ps1a search.py
Due Mon Sep 13.
- Read and report on SoM Chapter 4, The Self.
Due Fri Sep 10.
- Read and report on SoM Chapter 3, Conflict and Compromise.
Due Wed Sep 8.
- Read Society of Mind: Sections 2.1 through 2.6, on wholes and parts. Deliver synopses to me in one of three ways:
- 140-character tweet including reply-to tag
@fgmartand section number (e.g.,
2.1) [People following you but not following me won't see this.]
- 140-character tweet including hash tag
#umlaiand section number [All of your followers will see that you are thinking about SOM.]
- 140-character paragraphs in body of an email, with
umlaiin the subject line. Do not send attachments.
- 140-character tweet including reply-to tag
- ASSIGNMENT 1. Based on the in-class discussion, analyze a children's game or puzzle that is amenable to solving by a search process. Good examples of this are the game Rush Hour (which was handed out in class to a few people), peg solitaire games, such as the Cracker Barrel peg-jump game, and other one player board games.
- Initial state. Does the game have only one initial state? Does it have multiple possible initial states?
- Set of actions. Are these a function of the current state? Explain how.
- Transition model. Given a particular state and the set of possible actions for this state, is the transition to a new state deterministic? Explain.
- Goal test. Is there a definite end state? Can you tell if you won or lost and the game is over?
- Step cost. Do different actions have different costs? Or are all actions equal in cost, and you can model any move with a cost of 1?
- Download the Berkeley
search.zipcode, and make sure you can run it. Per the Discussion Group, the code may be run on
cs.uml.eduif your local machine has an X server.
python pacman.py --layout testMaze --pacman GoWestAgent.
- Continue reading AIMA sections 3.1 through 3.4.
Due Fri Sept 3.
- Read Society of Mind: Sections 1.1 through 1.6, on agents and agencies. Prepare 140-character synposes of each section. In each synopsis, include the hash tag
#umlaiand the decimal chapter/section number. You may either tweet each synopsis using your Twitter account, or email them (all synopses in one email message, please) to me (include
umlaiin the email subject line).
#umlai 1.1 Agents are the "particles that compose our minds." How do they interact, develop, gain awareness? This is what SOM will answer.
- Read AIMA Section 2: Intelligent Agents, which presents the intelligent agent framework that will permeate the presentation of material in the book. This material is mostly conceptual/architectural, rather than technical or mathematical.
- Read AIMA Section 3.1 through 3.4: Solving Problems by Searching, on so-called problem-solving agents that use search. Continue through the section on uninformed search (Section 3.4).
- Start looking at the Pac-Man Search assignment that we'll be doing.