when calling OptimizeHyperparameters to optimise my KNN model i am getting an error "out of memory".

on using the command

trainedClassifier=fitcknn(trainfeatures,trainlabels,'OptimizeHyperparameters','auto','HyperparameterOptimizationOptions',struct('AcquisitionFunctionName','expected-improvement-plus'))

i am getting the following error "Out of memory." after 5 iteration.

i am using '9.11.0.1873467 (R2021b) Update 3' version of matlab on 64 bit platform with RAM size 16GB.

Answer 1

[Srija Kethiri]

Hi Preeti,

I understand that you are getting an “Out of Memory” error after 5th iteration while using the “fitcknn” function.

Although the root cause of running out of memory is not clear, there are several ways to avoid this issue again. Here are options you can try which are not exclusive of each other, so feel free to try any combination of these:

1. The best one to start with is to increase swap space (also called virtual memory, which supplements physical RAM). In Windows 10, it is relatively easy to do so, follow the instructions on this link:          

      https://www.thewindowsclub.com/increase-page-file-size-virtual-memory-windows

2. Close all other unnecessary applications before running your script. Example: desktop applications like Chrome or IE will use a lot of memory when many tabs are open.

3. You can refer to the following documentation to know more about resolving the “out of memory” error issues:

https://in.mathworks.com/help/matlab/matlab_prog/resolving-out-of-memory-errors.html

Hope this helps!