Anonymous7 (Talk | contribs) |
|||
Line 14: | Line 14: | ||
---- | ---- | ||
− | ==Review== | + | =Review By Anonymous7= |
+ | |||
+ | |||
+ | |||
+ | [Review by Anoymous7] This slecture is about classification by using Bayes Rule in 1-dimensional and N-dimensional feature spaces. In the beginning, the author introduced Bayes theorem and gave 2 examples that use Bayes theorem. The author then discussed classification using Byes rule and derived the error formula for calculating the error when classifying 1 dimensional Gaussian distribution. After that, the author derived the discriminant function when classifying 1 dimensional and N dimensional Gaussian distribution. | ||
+ | |||
+ | <br>[Review by Anoymouse7] Overall, the slecture is very well written. The flow in the slecture seems to be smooth. However, to make the slecture better, the following improvements are suggested: | ||
+ | |||
+ | * The dimensional classification error is not written correct. Because of the author’s choice of classes <math> \omega_1 </math> and <math> \omega_2 </math>, the 1 dimensional classification error should be, | ||
+ | <center><math> | ||
+ | E(error) = \int_{-\infty}^{t}\rho(x|\omega_1)P(\omega_1)dx + \int_{t}^{\infty}\rho(x|\omega_2)P(\omega_2)dx | ||
+ | |||
+ | </math></center> | ||
+ | |||
+ | * In the Bayes rule example 1, it was written that P(W | L) = 0.75. It should be P(L | W) = 0.75 instead. | ||
+ | |||
+ | * When comparing the original probability and the probability that we get by applying Byes rule in examples 1 and 2, it should be explained why the probability changed. | ||
+ | * There should be a derivation of the discriminant function when it is the case of the general <math> \Sigma_i </math> in the N dimensional feature space. | ||
+ | |||
+ | |||
− | |||
---- | ---- |
Revision as of 20:35, 2 May 2014
Question and Comments for: Classification using Bayes Rule in 1-dimensional and N-dimensional feature spaces
A slecture by graduate student Jihwan Lee
Review By Anonymous7
[Review by Anoymous7] This slecture is about classification by using Bayes Rule in 1-dimensional and N-dimensional feature spaces. In the beginning, the author introduced Bayes theorem and gave 2 examples that use Bayes theorem. The author then discussed classification using Byes rule and derived the error formula for calculating the error when classifying 1 dimensional Gaussian distribution. After that, the author derived the discriminant function when classifying 1 dimensional and N dimensional Gaussian distribution.
[Review by Anoymouse7] Overall, the slecture is very well written. The flow in the slecture seems to be smooth. However, to make the slecture better, the following improvements are suggested:
- The dimensional classification error is not written correct. Because of the author’s choice of classes $ \omega_1 $ and $ \omega_2 $, the 1 dimensional classification error should be,
- In the Bayes rule example 1, it was written that P(W | L) = 0.75. It should be P(L | W) = 0.75 instead.
- When comparing the original probability and the probability that we get by applying Byes rule in examples 1 and 2, it should be explained why the probability changed.
- There should be a derivation of the discriminant function when it is the case of the general $ \Sigma_i $ in the N dimensional feature space.
Write other comment/question here.
Back to Classification using Bayes Rule in 1-dimensional and N-dimensional feature spaces