(New page: <center><font size= 4> Questions/Comments on slecture: Parzen Window Density Estimation </font size> A [https://www.projectrhea.org/learning/slectures...) |
|||
| (3 intermediate revisions by 2 users not shown) | |||
| Line 5: | Line 5: | ||
A [https://www.projectrhea.org/learning/slectures.php slecture] by [[ECE]] student Ben Foster | A [https://www.projectrhea.org/learning/slectures.php slecture] by [[ECE]] student Ben Foster | ||
| − | + | Partially based on the [[2014_Spring_ECE_662_Boutin|ECE662 Spring 2014 lecture]] material of [[user:mboutin|Prof. Mireille Boutin]]. | |
</center> | </center> | ||
---- | ---- | ||
| Line 11: | Line 11: | ||
| − | This is the talk page for the | + | This is the talk page for the slecture notes on [[Parzen Window Density Estimation|Parzen Window Density Estimation]]. Please leave me a comment below if you have any questions, if you notice any errors or if you would like to discuss a topic further. |
| Line 17: | Line 17: | ||
=Questions and Comments= | =Questions and Comments= | ||
| + | |||
| + | * This slecture has a very good and succinct introduction and overview of parametric and nonparametric density estimation methods, and clearly states why nonparametric density estimation methods are needed in practice. | ||
| + | * It also nicely shows the basic principles for Parzen Window method and explains the underlying assumptions and conditions behind the equation. | ||
| + | <center><math>p_n(\vec{X}) = \frac{k_n/n}{V_n}</math></center> | ||
| + | * It well described the experiment of the Parzen Window method with a hypercube window function with edge length <math>h</math>. And it showed how much the method can be sensitive to the choice of <math>h</math>. | ||
| + | ---- | ||
| + | * It could have been better if the slecture compared the Parzen Window method with different number of samples (training set sizes). | ||
| + | * Even though the slecture mentions about the popular Gaussian window function, there is no detailed discussion about the choice of different window functions and their density estimation results. | ||
| + | * Since we cannot visualize the higher dimension than 3D feature space, it can be improved by having classification results to compare different sample sizes and the parameter <math>h</math>. | ||
| + | ---- | ||
| + | |||
| + | Reviewed by Keehwan Park | ||
---- | ---- | ||
Latest revision as of 04:50, 6 May 2014
Questions/Comments on slecture: Parzen Window Density Estimation
A slecture by ECE student Ben Foster
Partially based on the ECE662 Spring 2014 lecture material of Prof. Mireille Boutin.
This is the talk page for the slecture notes on Parzen Window Density Estimation. Please leave me a comment below if you have any questions, if you notice any errors or if you would like to discuss a topic further.
Questions and Comments
- This slecture has a very good and succinct introduction and overview of parametric and nonparametric density estimation methods, and clearly states why nonparametric density estimation methods are needed in practice.
- It also nicely shows the basic principles for Parzen Window method and explains the underlying assumptions and conditions behind the equation.
- It well described the experiment of the Parzen Window method with a hypercube window function with edge length $ h $. And it showed how much the method can be sensitive to the choice of $ h $.
- It could have been better if the slecture compared the Parzen Window method with different number of samples (training set sizes).
- Even though the slecture mentions about the popular Gaussian window function, there is no detailed discussion about the choice of different window functions and their density estimation results.
- Since we cannot visualize the higher dimension than 3D feature space, it can be improved by having classification results to compare different sample sizes and the parameter $ h $.
Reviewed by Keehwan Park
