| Line 16: | Line 16: | ||
I have just a little suggestions about your slecture. | I have just a little suggestions about your slecture. | ||
| + | |||
It was not very clear how <math> h_j \longrightarrow0</math> results in<math> \frac{1}{V_j} \phi(\frac{x_l - x_0}{h_j}) = \delta_j(x_j - x_0) \longrightarrow \delta(x - x_0) </math>. | It was not very clear how <math> h_j \longrightarrow0</math> results in<math> \frac{1}{V_j} \phi(\frac{x_l - x_0}{h_j}) = \delta_j(x_j - x_0) \longrightarrow \delta(x - x_0) </math>. | ||
| + | |||
In addition, it would have been better if there was appropriate labels for each figure (x-axis was not very clear). Because the idea was on n dimensional while figures were 2D, figures were a little confusing. | In addition, it would have been better if there was appropriate labels for each figure (x-axis was not very clear). Because the idea was on n dimensional while figures were 2D, figures were a little confusing. | ||
| + | |||
If you would like to make the slecture even better, my suggestions is some graphs verifying the effects of smaller h or larger training data sets on density estimation and decision making based on Parzen windows method. | If you would like to make the slecture even better, my suggestions is some graphs verifying the effects of smaller h or larger training data sets on density estimation and decision making based on Parzen windows method. | ||
Revision as of 18:27, 4 May 2014
Comments for Parzen Windows
A slecture by Abdullah Alshaibani
Please leave me comment below if you have any questions, if you notice any errors or if you would like to discuss a topic further.
- Write comment/question here
This slecture is nicely organized from introduction to decision making based on Parzen windows. The slecture contains some important points from the class along with figures to help us understand better about window functions.
I have just a little suggestions about your slecture.
It was not very clear how $ h_j \longrightarrow0 $ results in$ \frac{1}{V_j} \phi(\frac{x_l - x_0}{h_j}) = \delta_j(x_j - x_0) \longrightarrow \delta(x - x_0) $.
In addition, it would have been better if there was appropriate labels for each figure (x-axis was not very clear). Because the idea was on n dimensional while figures were 2D, figures were a little confusing.
If you would like to make the slecture even better, my suggestions is some graphs verifying the effects of smaller h or larger training data sets on density estimation and decision making based on Parzen windows method.
But, overall, this slecture is excellent source to understand basic concepts of Parzen windows.
- answer here
