Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!

 



Forgot your password?
typodupeerror
×

Comment Re:really? (Score 1) 693

I agree that for some excellent students, it will be hard to claim that they cheated, but for a large portion of the cheaters, it will be easy. Remember, I'm not claiming that the only data used would be the scores from this test. Other sources of data: The fact that it's known there was a test bank, which allows us to claim that cheating occurred. Getting people to confess (see 'the prisoner's dilemma'). Analysis of friend groups. Past grades.

Comment Re:really? (Score 1) 693

Less useful, yes. Students with better records will have better cases. Students who decided to change their lives, do better in school and started with that test are screwed, though they might show different patterns of proficiency than the cheaters. Factors such as position on the tests bank pages, or overall order of test bank questions will influence the patterns of correct questions for the cheaters.

Once you have your 'suspects' are identified you work on the stronger cases via social pressure. As people confess, your algorithm gets more refined. The beauty of the problem is there are a large number of subjects, and a large number of cheaters. It would be fun detective work, except for the fact that that the situation is ethically depressing.

Comment Re:really? (Score 1) 693

On every test there are sets of questions that large amount of students get wrong, because of lack of emphasis in the classroom, or they are less obvious to study for. If there is an outside factor, like a test key, the students with the key will get these right. The more of these outlier questions they get right, the more likely it is that they cheated.

Comment Re:really? (Score 1) 693

There are a number of statistical approaches to determine who cheated. For instance, find people whose midterm grade is an anomaly compared to the rest of their grades. Next, look for particular patterns of questions that the cheaters got right, compared to those who didn't. Use a pattern matching algorithm to find to tease apart the bimodality of the grade distribution. There would be some students for which it is nearly certain that they cheated, and other for which it would be more uncertain. The students with higher average grades would have a better shot at arguing against having cheated, but the poor students would be sniffed out immediately.

Comment Re:Not the TSP (Score 2, Insightful) 394

Daylight is not what a bee is trying to conserve, it's flight distance. Bees minimize the distance flown to minimize the amount of energy they expend. The ratio they try to minimize is (energy expended)/(pollen collected). Pollen is turned into energy. When bees leave the hive they have a certain amount of energy they can expend. If a bee gets blown too far off track, or expends to much energy in some other way, it will run out of gas and die. But, it's better to see the problem from the perspective of the hive. The hive wants to gain as much energy as possible, while expending as little as possible.

So, actually the problem is fundamentally the same as TSP. It's a distance minimization problem. And just because they use a 'heuristic' doesn't mean that they don't have a solution to the TSP problem. An biologically-based genetic algorithm is no less valid than a computer algorithm.

Comment "The Calculus Lifesaver" (Score 1) 467

I was pretty much in the same boat you are. This book, and the accompanying videos, helped me to 'get my math back' after 15 years away. However, you might have to take a pre-calc refresher. It's amazing how much gets away from you after that much time.
Music

Submission + - Going Head To Head With Genius on Playlists (ucsd.edu)

brownerthanu writes: (This is a re-submit with the second paragraph revised to include an important part of the story.)

Engineers at the University of California, San Diego are developing a system to include an ignored sector of music, deemed the 'long tail', in music recommendations. It's well known that radio suffers from a popularity bias where the most popular songs receive an inordinate amount of exposure. In Apple's music recommender system, iTunes' Genius, this bias is magnified. An underground artist will never be recommended in a playlist due to insufficient data. It's an artifact of the popular collaborative filtering recommender algorithm, which Genius is based on.

In order to establish a more holistic model of the music world, Luke Barrington and researchers at the Computer Audition Laboratory have created a machine learning system which classifies songs in an automated, Pandora-like, fashion. Instead of using humans to explicitly categorize individual songs, they capture the wisdom of the crowds via a Facebook game, Herd It, and use the data to train statistical models. The machine can then "listen to", describe and recommend any song, popular or not. As more people play the game, the machines get smarter. Their experiments show that automatic recommendations work at least as well as Genius for recommending undiscovered music.

Music

Submission + - Going Head To Head With Genius on Playlists (ucsd.edu)

brownerthanu writes: Engineers at the University of California, San Diego are developing a system to include an ignored sector of music, deemed the 'long tail', in music recommendations. It's well known that radio suffers from a popularity bias, the most popular songs receive an inordinate amount of exposure. In Apple's music recommender system, iTunes' Genius, this bias is magnified. An underground artist will never be recommended in a playlist due to insufficient data. It's an artifact of the popular collaborative filtering recommender algorithm, which Genius is based on.

In order to establish a more holistic model of the music world, Luke Barrington, and researchers at The Computer Audition Laboratory have created a system which classifies songs in an automated, Pandora-like, fashion. Instead of using humans to explicitly categorize individual songs, they capture the wisdom of the crowds via a Facebook game, Herd It, and use the data to train statistical models. The algorithm can then "listen to", classify and recommend any song, popular or not.

Comment Brownerthanu (Score 1) 386

There are two other things I would test for:
  1. Generalization. There is experimental evidence which suggests that occupying perceptual resources creates a greater ability to generalize.
  2. Awareness of the distractor data. Who does better at gathering info from the distractor objects?

Comment chess AI (Score 1) 378

would something like this work? instead of crippling the AI, do enough move calculations so that the AI is guaranteed to blow almost any human opponent out of the water. rank the possible moves, and have the AI play one of the "less optimal" moves, depending on the chosen difficulty level.

Slashdot Top Deals

We are each entitled to our own opinion, but no one is entitled to his own facts. -- Patrick Moynihan

Working...