?url_ver=Z39.88-2004&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Adc&rft.title=Detection+of+Cognitive+States+from+fMRI+data+using+Machine+Learning+Techniques&rft.creator=Singh%2C+Vishwajeet&rft.creator=Miyapuram%2C+K.+P.&rft.creator=Bapi%2C+Raju+S.&rft.subject=Brain+Imaging&rft.subject=Machine+Learning&rft.description=Over+the+past+decade+functional+Magnetic+Resonance%0AImaging+(fMRI)+has+emerged+as+a+powerful%0Atechnique+to+locate+activity+of+human+brain+while%0Aengaged+in+a+particular+task+or+cognitive+state.+We%0Aconsider+the+inverse+problem+of+detecting+the+cognitive%0Astate+of+a+human+subject+based+on+the+fMRI%0Adata.+We+have+explored+classification+techniques%0Asuch+as+Gaussian+Naive+Bayes%2C+k-Nearest%0ANeighbour+and+Support+Vector+Machines.+In+order%0Ato+reduce+the+very+high+dimensional+fMRI+data%2C+we%0Ahave+used+three+feature+selection+strategies.+Discriminating%0Afeatures+and+activity+based+features%0Awere+used+to+select+features+for+the+problem+of%0Aidentifying+the+instantaneous+cognitive+state+given%0Aa+single+fMRI+scan+and+correlation+based+features%0Awere+used+when+fMRI+data+from+a+single+time+interval%0Awas+given.+A+case+study+of+visuo-motor+sequence%0Alearning+is+presented.+The+set+of+cognitive%0Astates+we+are+interested+in+detecting+are+whether+the%0Asubject+has+learnt+a+sequence%2C+and+if+the+subject+is%0Apaying+attention+only+towards+the+position+or+towards%0Aboth+the+color+and+position+of+the+visual%0Astimuli.+We+have+successfully+used+correlation%0Abased+features+to+detect+position-color+related+cognitive%0Astates+with+80%25+accuracy+and+the+cognitive%0Astates+related+to+learning+with+62.5%25+accuracy.&rft.date=2007&rft.type=Conference+Poster&rft.type=PeerReviewed&rft.format=application%2Fpdf&rft.identifier=http%3A%2F%2Fcogprints.org%2F5364%2F1%2FIJCAI07-093.pdf&rft.identifier=++Singh%2C+Vishwajeet+and+Miyapuram%2C+K.+P.+and+Bapi%2C+Raju+S.++(2007)+Detection+of+Cognitive+States+from+fMRI+data+using+Machine+Learning+Techniques.++%5BConference+Poster%5D+++++&rft.relation=http%3A%2F%2Fcogprints.org%2F5364%2F