
USAP PAETE Discussion Forums for the people of Paete, Laguna, Philippines

View previous topic :: View next topic 
Author 
Message 
adedios SuperPoster
Joined: 06 Jul 2005 Posts: 5060 Location: Angel C. de Dios

Posted: Tue Jan 31, 2006 7:49 am Post subject: (Math) Patterns: Ways To Help Computers Recognize Patterns 


Source: Ohio State University
Date: 20060131
URL: http://www.sciencedaily.com/re.....154338.htm

A New Way To Help Computers Recognize Patterns
Researchers at Ohio State University have found a way to boost the development of pattern recognition software by taking a different approach from that used by most experts in the field.
This work may impact research in areas as diverse as genetics, economics, climate modeling, and neuroscience.
Aleix Martinez, assistant professor of electrical and computer engineering at Ohio State, explained what all these areas of research have in common: pattern recognition.
He designs computer algorithms to replicate human vision, so he studies the patterns in shape and color that help us recognize objects, from apples to friendly faces. But much of today's research in other areas comes down to finding patterns in data  identifying the common factors among people who develop a certain disease, for example.
In fact, the majority of pattern recognition algorithms in science and engineering today are derived from the same basic equation and employ the same methods, collectively called linear feature extraction, Martinez said.
But the typical methods don't always give researchers the answers they want. That's why Martinez has developed a fast and easy test to find out in advance which algorithms are best in a particular circumstance.
"You can spend hours or weeks exploring a particular method, just to find out that it doesn't work," he said. "Or you could use our test and find out right away if you shouldn't waste your time with a particular approach."
The research grew out of the frustration that Martinez and his colleagues felt in the university's Computational Biology and Cognitive Science Laboratory, when linear algorithms worked well in some applications, but not others.
In the journal IEEE Transactions on Pattern Analysis and Machine Intelligence, he and doctoral student Manil Zhu described the test they developed, which rates how well a particular pattern recognition algorithm will work for a given application.
Along the way, they discovered what happens to scientific data when researchers use a lessthanideal algorithm: They don't necessarily get the wrong answer, but they do get unnecessary information along with the answer, which adds to the problem.
He gave an example.
"Let's say you are trying to understand why some patients have a disease. And you have certain variables, which could be the type of food they eat, what they drink, amount of exercise they take, and where they live. And you want to find out which variables are most important to their developing that disease. You may run an algorithm and find that two variables  say, the amount of exercise and where they live  most influence whether they get the disease. But it may turn out that one of those variables is not necessary. So your answer isn't totally wrong, but a smaller set of variables would have worked better," he said. "The problem is that such errors may contribute to the incorrect classification of future observations."
Martinez and Zhu tested machine vision algorithms using two databases, one of objects such as apples and pears, and another database of faces with different expressions. The two tasks  sorting objects and identifying expressions  are sufficiently different that an algorithm could potentially be good at doing one but not at the other.
The test rates algorithms on a scale from zero to one. The closer the score is to zero, the better the algorithm.
The test worked: An algorithm that received a score of 0.2 for sorting faces was right 98 percent of the time. That same algorithm scored 0.34 for sorting objects, and was right only 70 percent of the time when performing that task. Another algorithm scored 0.68 and sorted objects correctly only 33 percent of the time.
"So a score like 0.68 means 'don't waste your time,'" Martinez said. "You don't have to go to the trouble to run it and find out that it's wrong twothirds of the time."
He hopes that researchers across a broad range of disciplines will try out this new test. His team has already started using it to optimize the algorithms they use to study language and cancer genetics.
This work was sponsored by the National Institutes of Health.
*************************************************************
Questions to explore further this topic:
Given the following images, could you recognize what emotions may be behind these faces?
http://rvl1.ecn.purdue.edu/~aleix/faceex2.html
How does one recognize these emotions from the faces?
http://researchnews.osu.edu/archive/compvisn.htm
How can machines understand human gestures?
http://vismod.media.mit.edu/sa.....esture.mpg
Can a machine interact with people?
http://robotic.media.mit.edu/p.....robot.html
Can a machine achieve social learning?
http://robotic.media.mit.edu/p.....rning.html
http://robotic.media.mit.edu/p.....obots.html
Do you recognize a pattern in the picture below?
We are all smiling!
Here are drugs used against malaria, do you recognize a pattern?
What is pattern recognition?
http://www.learner.org/channel....._03_b.html
What are Venn diagrams?
http://www.learner.org/channel....._04_a.html
http://regentsprep.org/Regents/math/venn/LVenn.htm
http://www.shodor.org/interact...../vdiagram/
http://www.shodor.org/interact.....s/venndia/
http://www.combinatorics.org/S.....nnEJC.html
http://www.sdcoe.k12.ca.us/score/actbank/tvenn.htm
http://www.graphic.org/venbas.html
http://www.readwritethink.org/materials/venn/
Now, Let us see if you can recognize patterns:
http://www.learner.org/channel....._01_a.html
Is recognition important in our learning?
http://www.cast.org/publicatio.....pter2.html
What is machine learning?
http://www.cs.wisc.edu/%7Edyer.....rning.html
A book on machine learning
http://ai.stanford.edu/people/nilsson/mlbook.html
A complete course on machine learning and pattern recognition
http://www.kddresearch.org/Cou...../Lectures/
GAMES
http://pbskids.org/sesame/ernie/e_surprisebox.html
http://pbskids.org/sesame/ernie/e_cluehunt.html
http://vismod.media.mit.edu/vi.....story.html
Last edited by adedios on Sat Jan 27, 2007 3:10 pm; edited 2 times in total 

Back to top 


TonyB
Joined: 05 Jan 2006 Posts: 178

Posted: Tue Jan 31, 2006 10:07 am Post subject: Most difficult 


Pattern recognition is one of the most difficult to do programmingwise. If this obstacle is hurdled, we are looking at great improvements in many areas that up to now still require humans to do. Take for example, a bad one  the antispam mechanism where site registrations require that you type the funnylooking letters and numbers displayed. They are still to be overcome by robots that automatically generate spam. (Thank goodness.)
As a student, I took 3 units of artificial intelligence or AI because I was curious but I was also diappointed because it didn't offer me anything that's really new to me then. (I did advanced reading.) Pattern recognition then (10 years ago) was just a hodgepodge of theories on how problems could be approached. Up to now, it still have to progress into what is "acceptable" so that it can break into the mainstream and into consumer products. (The one thing we are most familiar with are barcode readers.) We already see working applications using speech recognizion, but reliable visual ID is still far from being a "household" name. Part of the reason is that visual requires enormous amount of computing power for it to become practical reality. We are getting there with current hardware but software is still lagging behind.
cheers!
tony.basa _________________ Usap Business  LDP Batch '89  This Side Of Town 

Back to top 


adedios SuperPoster
Joined: 06 Jul 2005 Posts: 5060 Location: Angel C. de Dios

Posted: Mon Nov 27, 2006 11:12 am Post subject: Computer Detects Anger Before Fights Break Out 


Computer Detects Anger Before Fights Break Out
By Bill Christensen
posted: 27 November 2006
07:57 am ET
You can tell when you hear someone is really angry—and get out of the way. Now Sigard, a new software package developed by Sound Intelligence, can also detect verbal aggression with a high level of accuracy.
Combined with closed circuit television systems, Sigard can quickly notify security personnel about loud, angry people in outdoor public spaces, public transportation, nightclubs and bars.
Here's how it works. A single analysis computer accepts sensor input from a variety of locations. Once the software detects a verbally aggressive human voice, it activates the camera associated with that sensor, bringing it to a security guard's attention. This helps cut down on the number of people needed to monitor CCTVs.
For the full article:
http://www.livescience.com/sci.....sound.html 

Back to top 


adedios SuperPoster
Joined: 06 Jul 2005 Posts: 5060 Location: Angel C. de Dios

Posted: Sun Dec 03, 2006 7:09 am Post subject: Artistic Artificial Life 


Artistic Artificial Life
2 December 2006
Explore how patterns are created on a computer. This Webbased project represents the work of three Calgary artists: Vera Gartley, Arlene Stamp, and Mary Shannon Will.
http://www.artificiallife.net/ 

Back to top 


adedios SuperPoster
Joined: 06 Jul 2005 Posts: 5060 Location: Angel C. de Dios

Posted: Fri Apr 13, 2007 9:50 am Post subject: Feds: Accuracy of Face Recognition Software Skyrockets 


Feds: Accuracy of Face Recognition Software Skyrockets
By Lamont Wood
Special to LiveScience
posted: 13 April 2007
08:57 am ET
After months of testing, federal investigators interested in gauging its usefulness in national security applications say face recognition software has improved by a factor of 20 in the last five years.
At least that's the recently reported results of the latest Face Recognition Vendor Test (FRVT) conducted during the past year by the National Institute of Standards and Technology (NIST) to determine which algorithms did the best job of verifying a person's identity by examining his or her face.
Jonathon Phillips, an electrical engineer who directed the test at NIST, explained to LiveScience that a similar FRVT conducted in 2002 showed that the best algorithms failed to make a correct comparison 20 percent of the time. But the rate of false rejections in the latest test was only 1 percent.
For the full article:
http://www.livescience.com/tec.....ition.html 

Back to top 


adedios SuperPoster
Joined: 06 Jul 2005 Posts: 5060 Location: Angel C. de Dios

Posted: Sat May 05, 2007 8:10 am Post subject: The Mathematical Lives of Plants 


Week of May 5, 2007; Vol. 171, No. 18
The Mathematical Lives of Plants
Julie J. Rehmeyer
The seeds of a sunflower, the spines of a cactus, and the bracts of a pine cone all grow in whirling spiral patterns. Remarkable for their complexity and beauty, they also show consistent mathematical patterns that scientists have been striving to understand.
For the full article:
http://sciencenews.org/article.....thtrek.asp 

Back to top 


adedios SuperPoster
Joined: 06 Jul 2005 Posts: 5060 Location: Angel C. de Dios

Posted: Fri May 11, 2007 6:45 am Post subject: Americans and Japanese Read Faces Differently 


Americans and Japanese Read Faces Differently
By Melinda Wenner, Special to LiveScience
posted: 10 May 2007 08:36 am ET
Culture is a huge factor in determining whether we look someone in the eye or the kisser to interpret facial expressions, according to a new study.
For instance, in Japan, people tend to look to the eyes for emotional cues, whereas Americans tend to look to the mouth, says researcher Masaki Yuki, a behavioral scientist at Hokkaido University in Japan.
This could be because the Japanese, when in the presence of others, try to suppress their emotions more than Americans do, he said.
In any case, the eyes are more difficult to control than the mouth, he said, so they probably provide better clues about a person's emotional state even if he or she is trying to hide it.
For the full article:
http://www.livescience.com/hea.....lture.html 

Back to top 


adedios SuperPoster
Joined: 06 Jul 2005 Posts: 5060 Location: Angel C. de Dios

Posted: Mon May 21, 2007 8:23 pm Post subject: Face Recognition: Bobs Don't Look Like Tims 


Face Recognition: Bobs Don't Look Like Tims
By Melinda Wenner, Special to LiveScience
posted: 21 May 2007 11:46 am ET
It's easier to remember a "Bill" who really fits the bill, according to a new study.
Names tend to be associated with certain facial featuresBobs have rounder faces than Tims, for exampleand it's easier to learn a person's name if his face matches it.
Robin Thomas, a cognitive scientist at Miami University in Ohio, noticed that she frequently confused the names of two of her students. This didn't happen to her often, so she wondered if there was more to it than just forgetfulness.
Then she realized this. "Their faces did not fit the name they were given," Thomas said.
For the full article:
http://www.livescience.com/hea.....ition.html 

Back to top 


adedios SuperPoster
Joined: 06 Jul 2005 Posts: 5060 Location: Angel C. de Dios

Posted: Sun Sep 02, 2007 6:12 am Post subject: The Wealth of Nations 


Week of Sept. 1, 2007; Vol. 172, No. 9 , p. 138
The Wealth of Nations
A country's competitive edge can spread industry to industry, like a disease
Davide Castelvecchi
The economies of poor and developing countries often depend almost exclusively on a single product—perhaps timber or coffee—or on a handful of products at most. That's hardly a startling observation, but what's puzzled economists over the years is why it's been so difficult for these countries to start up new activities in the hope of spurring economic growth and lifting themselves out of poverty.
For the full article:
http://sciencenews.org/articles/20070901/bob9.asp 

Back to top 


adedios SuperPoster
Joined: 06 Jul 2005 Posts: 5060 Location: Angel C. de Dios

Posted: Sat Sep 29, 2007 8:04 am Post subject: Beating the Bush for Patterns 


Week of Sept. 29, 2007; Vol. 172, No. 13
Beating the Bush for Patterns
Landscapes lacking a fractal vegetation pattern may be vulnerable to desertification.
Julie J. Rehmeyer
In Africa's Kalahari Desert as well as some areas around the Mediterranean, trees and bushes grow in clumps scattered in seemingly random locations across an otherwise barren landscape. Two new studies have discovered a fractal pattern in this seeming randomness, and they offer a novel explanation of how it comes about. One study suggests that areas without such a pattern are on the edge of collapse, and that further pressure could tip them into becoming barren deserts.
For the full article:
http://sciencenews.org/article.....thtrek.asp 

Back to top 


adedios SuperPoster
Joined: 06 Jul 2005 Posts: 5060 Location: Angel C. de Dios

Posted: Fri Jan 04, 2008 2:36 pm Post subject: Modular counting: Of election dates and other patterns 


Modular counting: Of election dates and other patterns
STAR SCIENCE By Carlene P.C. PilarArceo, PhD
Thursday, January 3, 2008
Since the 1995 elections I’ve been involved in election campaigns. Preferring to stay behind the scenes, I invariably started out doing desk and paper work, like scheduling, data banking and bookkeeping. To put color in that sixweek daily grind, I volunteered to compose campaign jingles as well.
My candidate then was a municipal councilor newly elected in 1992 and going for a second second term in 1995. In updating the campaign jingle he used in 1992, I remember wondering if we could just edit the voiceover, part of which exhorted the listeners to vote for soandso candidate on election day. That went something like, “Mga kababayan, pagdating ng Mayo onse, iboto si…” My initial idea was to simply replace the date “onse” with whatever date the day of elections would fall on, like “otso” or “katorse,” and to replace “iboto” with “ibalik.” These days, with accessible high technology, that could have easily been done. A dozen years ago, armed with just cassette tapes, it was an entirely different story. Come jingleplaying time, one’s best option was a tape deck which could do autoreverse.
National and local elections fall on the second Monday of the month of May. In 1992, election day fell on May 11; in 1995, on May 14; in 1998, on May 11. Just when I was beginning to think that we could save campaign money in a thirdterm campaign in 1998 by either deploying the 1992 jingle again or, technology now allowing, replacing “katorse” with “onse” in the 1995 jingle, my candidate decided to run for the position of municipal mayor instead. But let’s not go into my candidate’s political forays. Let’s, you and me, go instead on a mathematical foray.
I will not discuss the mathematics found in a campaign’s budgeting, bookkeeping and forecasting. The budgeting is simple arithmetic, the bookkeeping is basic accounting, and the forecasting can be easily rendered futile or inutile by the intricacies and convolutions of campaign dynamics. Through this article I intend to introduce you to modular counting.
Modular counting is best described by the operation – division. Modular counting assigns to a number the remainder when that number is divided by a given fixed number, called the modulo or simply, mod. Consider the positive integers:
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 …
If I divide each of the numbers above by 2, the remainder is either 0 or 1, respectively, depending on whether the number is even or odd. Assigning to each number above the remainder when it is divided by 2, we get
1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 …
Hence, we say 1 mod 2 = 1, 2 mod 2 = 0, 3 mod 2 = 1, 4 mod 2 = 0, and so on. Evidently, odd numbers are assigned the value 1 and even numbers are assigned the value 0 in mod 2 counting. The alternation of values is crucial; it is impossible to have two consecutive 0’s or 1’s – this means having two consecutive even numbers or two consecutive odd numbers! Think of the cadence of a march, and drill commanders shouting “Left! Right! Left! ...” Is it not impossible to march properly to “Left! Left! Left! …”? It is likewise impossible to turn a switch on and then on again; it must be switched off in between. Even for relationships, is not the term “onagain, offagain” and not “onagain, onagain”?
Let’s take a look at more illustrations. Consider mod 3 counting. This means we divide each of the positive integers listed above by 3 and assign the respective remainders:
1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 …
Assigned “1” are 1, 4, 7, and so on. Assigned “2” are 2, 5, 8, and so on. Assigned “0” are 3, 6, 9, and so on. Those who play mahjong will recognize the numbers in a puro in the upanddown or escalera configurations, where the aim is the completion of consecutive trios. Nonmahjong players may more clearly understand mod 3 counting in the setup of a traffic light: green, yellow, red, or equivalently, go, slow down, stop. Needless to say, the sequences in mahjong and traffic lights can never be jumbled.
For mod 4 counting, our country’s fondness for beauty pageants makes me choose quarter turns as a good example. Unlike military drills which can make cadets/ettes turn right or left at random, a beauty pageant standard quarter turn is usually done to one’s left, and woe is the candidate who turns right! Being quarter turns, these are done four times until the candidates face front again. I am not sure whether candidates are still made to do these, but obviously these allow judges to view candidates from all sides – front and back, left and right. Just as inappropriate as a candidate turning right would be any bungling of the sequence of numbers in mod 4 counting. Observe, division of the positive integers above by 4 gives:
1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 …
Of course, 0, 1, 2 and 3 are the only possible remainders when dividing by 4. Equivalently, “1” as the first quarter turn has the candidate facing her left, “2” has the candidate facing back, “3” has the candidate facing her right, and “0” returns the candidate to the original position.
Here are some more familiar and everyday examples of modular counting. For each example verify that elements of the sequences can never be arranged differently.
Mod 7 is seen in the days of the week, taking Monday as 1, Tuesday as 2, and so on until Saturday as 6 and Sunday as 0. It is also represented by notes on a scale, with Do as 1, Re as 2, Mi as 3, and so on until Ti as 7, “which will bring us back to Do” as 0, as Julie Andrews sang in The Sound of Music.
Mod 12 is easily illustrated by the months of a year, with January as 1 and December as 0, and the numbers on a clock or watch face, with 12:00 as 0. In fact, most digital timepieces already reflect 12:00 midnight as 00:00. Military time will give us mod 24 counting.
Mod 26 is already being practiced by millions as early as high school, albeit unwittingly. Consider the English alphabet and assign 1 to letter A, 2 to letter B, and so on until 25 to letter Y and 26 (or 0) to letter Z. A high school student who accidentally bites her tongue excitedly asks her girlfriends for a number and immediately converts this to a letter corresponding to that number, according to the assignment mentioned. For example, the number 18 would correspond to the letter R, and the girlfriends would squeal, “Uuy, naalala siya ni Rodney!” Or, the number 28, in mod 26, would correspond to the letter B, and point to a Bobet. Sometimes, bigger numbers are requested so that the correspondence is less obvious (and intentional). Hence, the number 81 would correspond to the letter C and be interpreted as Caloy, and number 105 would correspond to the letter A and be interpreted as Ariel. That’s kilig care of mod 26!
Then there was the big Y2K scare at the turn of the millennium when all programs were only configured until year 1999, so that at midnight of December 31, 1999, all settings for “year” will go to 0. That’s mod 2000 counting!
Let’s slow down a bit and go to mod 28 and mod 84. Why these modulo’s? Because in my mind’s meanderings about election dates, my paper and pencil gave me these numbers of years for the second Mondays of May to completely repeat their pattern of dates.
Year  Date of 2nd Monday of May
1989* 8
1992 11
1995 8
1998 11
2001 14
2004 10
2007 14
2010 10
2013 13
2016 9
2019 13
2022 9
2025 12
2028 8
2031 12
2034 8
2037 11
2040 14
2043 11
2046 14
2049 10
2052 13
2055 10
2058 13
2061 9
2064 12
2067 9
2070 12
*assuming elections were held this year
Indeed, it’s not as obvious as our previous examples. We can play this in two ways. First, we may assign the number 1 to 1989, 2 to 1992, and so on until 27 for 2067 and 0 for 2070 and come up with mod 28. More accurately, we assign the number 1 to 1989, 2 to 1990, 3 to 1991, 4 to 1992, and so on until 83 for 2069 and 0 for 2070 and come up with mod 84. If you have the time and inclination, I encourage you not only to fill in the blanks in the table above, but better yet to expand the table and include the years in between, like 1990 and 1991, to be able to see the entire list and pattern. Of course, expect only to find entries ranging from 8 to 14, as these are the only possible dates for a second Monday (or any second — day, for that matter).
For mod 28 also, we can play a birthday game. Find out on what particular day you were born and verify that your 28th birthday will fall, or fell, on the same day. Of course it can happen before then, like my day of birth was the same for my fifth, 11th and 22nd birthdays, but the pattern of days your birthday will fall on (for all years from the first through the 27th), will repeat completely beginning on your 28th birthday, as if you’re a newborn again and back to age 0, a la mod 28. This is exactly the essence of modular counting.
For the true numerophile, or the brave and daring, I recommend mod 3300. This is the number of years that our calendar year will be in synch with the solar year, or the time it takes the Earth to complete its orbit around the Sun. A calendar year is actually about a quarterday short of a solar year, and leap years are meant to take care of that. All the calculations involved – think mod 4, mod 100 and mod 400 – will show that it will take 3,300 years before the calendar and solar years will again be out of synch (look up websites such as infoplease.com or wikipedia.org).
You may question the utility of this, since we most probably won’t be around anymore for the turn of the century, more so the next millennium. I do not need it anymore as a campaign worker. I have since given up on fundsaving efforts with respect to campaign jingles. You see, just last May my candidate ran for provincial board member so coming up with a campaign jingle has become much less simple than a mere change of dates in the voice over. But hey, you can always ask your child to do mod 3300 counting and, lo and behold, put him or her to sleep with minimum effort. Now isn’t that a good thing?
* * *
Cayen Arceo is an associate professor of Mathematics in UP Diliman. Her research areas include partial differential equations and operations research. A current interest is General Education Mathematics which she has been teaching for the past several semesters. 

Back to top 




You can post new topics in this forum You can reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum

Powered by phpBB © 2001, 2005 phpBB Group
