In my research work i need to find out those features which are most significant for classification. Or I need to find out most useful features and again i have to train classifier and need to check accuracy with reduced feature. [all these i need to do with pso] so my question is how I can obtain reduced features?
I have written following code for PSO update of postion and velocity
for i in range(rows): # rows = no of samples
# dim = 30 which is randomly selected from 384 dimensions
for j in dim:
Hi,
As matter of fact, ANN or nonlinear system identification is suitable approaches to select optimized features. I have done some works about optimized features. you could catch a glimpse of them.
In our paper Article Simultaneous feature selection and parameter optimisation of...
we have inspired from this code https://www.mathworks.com/matlabcentral/fileexchange/46961-binary-genetic-algorithm-feature-selection-zip?focused=7733492&tab=function to implement PSO for feature selection while training the SVM model.
Respected Dr. Suman Tuwari,
Hi,
Your appropriate reponse should be found in this link:
Article Heuristic optimization-based wave kernel descriptor for defo...
This is exactly that you are looking for:
| * | * | * | * | * | * | * | * | * | * | * | * | * |
V V V V V V V V V V V V V V
We proposed an optimized descriptor method for the 3D shape matching and retrieval. The advantage of the improved method comes from the tuning of a parameter of a kernel descriptor using a stochastic method called PSO. By this way, we seek for a descriptor which should be as discriminate as possible between 3D shapes. Hence, the superiority of the method is direct, since it uses parameters that have been optimized. In other words, the new descriptor is excellent because it is more finely tuned for a precise and fixed database and especially for each class of object.
Hope this helps.
Best regards
Hi,
As matter of fact, ANN or nonlinear system identification is suitable approaches to select optimized features. I have done some works about optimized features. you could catch a glimpse of them.
@Mohammad : can you suggests the resources from where i can read and can go ahead. I have implemented PSO for feature optimization but I am not getting how to get those optimized feature's name.
tyr the following paper:
Article Instance based dynamic feature selection for network traffic...
- Badges
- Science topic
- Similar topics
- Organizational Psychology
- Training