I am working on KFCM for clustering of dataset. The output of it centroids, membership matrix and objective function.
I referred paper "Identification of flight state under different simulator modes using diffusion maps" and found that optimization of KFCM paramaeters can be done using hybrid algorithm for better accuracy.
Can anyone tell me what is the meaning of optimization of KFCM parameters?
Hello,
Actually the k-FCM is a clustering algorithm using the distance between the k centroids and your data including a degree of membership in the cost function.
Then by an iterative process, the distance is computed by changing the location of yours centroids, the assignment of your data to centroids and the degree of membership.
The optimization work in this process by selecting the parameters at the step i-1 and creating new parameter at the step i. And the hybrid algorithm is actually a combination of least-squares estimation and back-propagation. It is the algorithm used by default.
I hope these information help you
Thanks .I got good information regarding KFCM. But I am unable to get the optimization process explained by you. Parameter in i-1 and new parameter in i. Here Hybrid PSO-BFO is used. What can be the parameters and how optimization is done?
- Badges
- Science topic
- Similar topics
- Computer Communications (Networks)
- Simulators