The current implementation of the computeKernelBasisVectors function assumes that every class or subset of samples in the dataset will have an equal number of samples. This assumption might not hold in many real-world datasets, where class distributions could be imbalanced.
Expected Behavior:
The function should dynamically adjust based on the actual number of samples in each subset or class, without making assumptions about uniform class distributions.
Current Behavior:
The function assumes that every class the dataset, represented by the 4rd-order of the multi-dimensional matrix X, has the same number of samples.
Steps to Reproduce:
Use a dataset with varying numbers of samples in different subsets or classes.
Attempt to compute basis vectors using the computeKernelBasisVectors function.
Possible Solution:
Refactor the function to account for different sample sizes in classes Consider using dynamic data structures such as cell arrays or loops that account for the actual number of samples in each subset rather than a fixed size.
Additional Information:
This issue limits the flexibility of the function when working with imbalanced datasets. Addressing this issue will enhance the robustness and flexibility of the code.
Santos-Enoque
commented
1 year ago 
Issue Description:
The current implementation of the
computeKernelBasisVectorsfunction assumes that every class or subset of samples in the dataset will have an equal number of samples. This assumption might not hold in many real-world datasets, where class distributions could be imbalanced.Expected Behavior:
The function should dynamically adjust based on the actual number of samples in each subset or class, without making assumptions about uniform class distributions.
Current Behavior:
The function assumes that every class the dataset, represented by the 4rd-order of the multi-dimensional matrix
X, has the same number of samples.Steps to Reproduce:
computeKernelBasisVectorsfunction.Possible Solution:
Refactor the function to account for different sample sizes in classes Consider using dynamic data structures such as
cellarrays or loops that account for the actual number of samples in each subset rather than a fixed size.Additional Information:
This issue limits the flexibility of the function when working with imbalanced datasets. Addressing this issue will enhance the robustness and flexibility of the code.