Cross-Validation Accuracy very high (>90%) at random signal in BCI motor imagery CSP Scenario

Working with OpenViBE signal processing scenarios and doing scenario/BCI design
Post Reply
seidi
Posts: 1
Joined: Mon Jan 18, 2021 3:45 pm

Cross-Validation Accuracy very high (>90%) at random signal in BCI motor imagery CSP Scenario

Post by seidi »

Hi,

I was getting chance level accuracy for a BCI motor imagery scenario on subjects, so I tried to run this scenario without a subject and see what results it gives.
It happens that cross-validation accuracy was really high multiple times (CSP + LDA as processing pipeline) with EEG cap on the table. Any recommendations on how to debug this problem?

Thanks!

I can provide screenshots and more info if needed. I analysed this data on python and it gave chance level accuracy, fortunately

Post Reply