| Line 1: | Line 1: | ||
| − | + | [[Category:ECE662Spring2012Boutin]] | |
| − | + | ||
| − | [[Category: | + | |
[[Category:blog]] | [[Category:blog]] | ||
| Line 7: | Line 5: | ||
Tuesday January 10, 2012 (Week 1) | Tuesday January 10, 2012 (Week 1) | ||
---- | ---- | ||
| − | We began the lecture by going over the [[Media:syllabusECE662Boutin.pdf|syllabus]]. We then defined "pattern recognition" and went over a few examples of pattern recognition problems. The emphasis was put on image processing related problems, as many students in the class are working in image processing. We noted that in pattern recognition, one always chooses among a 'finite' set of classes, labeled 1,2,... n. (However, in order to make such a choice, we may need to estimate continuous-valued functions or a set of real-valued parameters, as we will see later.) We used a toy problem to illustrate the statistical pattern recognition paradigm. In this toy problem, a game show host is asking a contestant to guess the gender of a person hidden behind a curtain. The strategy we used was to try to optimize the | + | We began the lecture by going over the [[Media:syllabusECE662Boutin.pdf|syllabus]]. We then defined "pattern recognition" and went over a few examples of pattern recognition problems. The emphasis was put on image processing related problems, as many students in the class are working in image processing. We noted that in pattern recognition, one always chooses among a 'finite' set of classes, labeled 1,2,... n. (However, in order to make such a choice, we may need to estimate continuous-valued functions or a set of real-valued parameters, as we will see later.) |
| + | |||
| + | We used a toy problem to illustrate the statistical pattern recognition paradigm. In this toy problem, a game show host is asking a contestant to guess the gender of a person hidden behind a curtain. The strategy we used was to try to optimize the chance of being right. Without any information, we had a 50% chance of being right by guessing "male". This percentage increased to 90% when the fact that the person was a Purdue [[ECE]] person was revealed. However, when the hair length of the person was revealed (30 cm), we decided to change our guess based on the information that only 1 male in [[ECE]] has 30 cm long hair, compared with 5 females. | ||
| + | |||
| + | We ended the lecture with a simple question: based on the information given (Purdue ECE and 30 cm long hair), would it be better to guess randomly among male or female with a 1:5 probability, or is it better to always stick with the most likely outcome, namely female? You can learn more on [[EE662Sp10OptimalPrediction|this page]] (created by a student in 2010). | ||
=Action Items= | =Action Items= | ||
| Line 16: | Line 18: | ||
*[[Media:Capcha.pdf|Example of recognition problems as CAPCHAs]] | *[[Media:Capcha.pdf|Example of recognition problems as CAPCHAs]] | ||
*[[What_is_Pattern_Recognition_OldKiwi|"What is pattern recognition?" (class notes from 2010, written by students)]] | *[[What_is_Pattern_Recognition_OldKiwi|"What is pattern recognition?" (class notes from 2010, written by students)]] | ||
| + | *[[EE662Sp10OptimalPrediction|Page discussing difference between choosing most likely outcome always, or randomly guessing following the relative probabilities of the outcome]] | ||
*[[ECE662:Glossary_Old_Kiwi|Pattern Recognition Glossary (written by ECE662 students in 2010)]] | *[[ECE662:Glossary_Old_Kiwi|Pattern Recognition Glossary (written by ECE662 students in 2010)]] | ||
*[[Help:Contents|Rhea Help Page]] | *[[Help:Contents|Rhea Help Page]] | ||
Next: [[Lecture2ECE438F11|Lecture 2]] | Next: [[Lecture2ECE438F11|Lecture 2]] | ||
| + | ---- | ||
| + | ==Comments== | ||
| + | Please write your comments and questions below. | ||
| + | *Write a comment here | ||
| + | *Write another comment here. | ||
| + | |||
---- | ---- | ||
[[2011 Fall ECE 438 Boutin|Back to 2011 Fall ECE 438 Boutin]] | [[2011 Fall ECE 438 Boutin|Back to 2011 Fall ECE 438 Boutin]] | ||
Revision as of 08:41, 10 January 2012
Contents
Lecture 1 Blog, ECE662 Spring 2012, Prof. Boutin
Tuesday January 10, 2012 (Week 1)
We began the lecture by going over the syllabus. We then defined "pattern recognition" and went over a few examples of pattern recognition problems. The emphasis was put on image processing related problems, as many students in the class are working in image processing. We noted that in pattern recognition, one always chooses among a 'finite' set of classes, labeled 1,2,... n. (However, in order to make such a choice, we may need to estimate continuous-valued functions or a set of real-valued parameters, as we will see later.)
We used a toy problem to illustrate the statistical pattern recognition paradigm. In this toy problem, a game show host is asking a contestant to guess the gender of a person hidden behind a curtain. The strategy we used was to try to optimize the chance of being right. Without any information, we had a 50% chance of being right by guessing "male". This percentage increased to 90% when the fact that the person was a Purdue ECE person was revealed. However, when the hair length of the person was revealed (30 cm), we decided to change our guess based on the information that only 1 male in ECE has 30 cm long hair, compared with 5 females.
We ended the lecture with a simple question: based on the information given (Purdue ECE and 30 cm long hair), would it be better to guess randomly among male or female with a 1:5 probability, or is it better to always stick with the most likely outcome, namely female? You can learn more on this page (created by a student in 2010).
Action Items
Get familiar with Rhea. Try posting something somewhere.
Relevant Material
- course syllabus.
- Example of recognition problems as CAPCHAs
- "What is pattern recognition?" (class notes from 2010, written by students)
- Page discussing difference between choosing most likely outcome always, or randomly guessing following the relative probabilities of the outcome
- Pattern Recognition Glossary (written by ECE662 students in 2010)
- Rhea Help Page
Next: Lecture 2
Comments
Please write your comments and questions below.
- Write a comment here
- Write another comment here.
