BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Quantifying the applicability domain between two datasets using Ta
 nimoto similarity - Marcus Wang\, University of Cambridge
DTSTART:20210210T143000Z
DTEND:20210210T150000Z
UID:TALK153676@talks.cam.ac.uk
CONTACT:Lisa Masters
DESCRIPTION:Machine learning is a popular technique used in predictive tox
 icology\, where predicting the activity of compounds on targets/endpoints 
 is an important aspect. During the process of machine learning\, training 
 and test datasets need to be used where the applicability domain between t
 hese two datasets would give an indication of the performance of the machi
 ne learning model\, as well as whether the model can be applied to other d
 atasets. In this work\, a method that uses Tanimoto similarity on molecula
 r fingerprints is described that quantifies the applicability domain betwe
 en two datasets. A total of 76 human targets\, of which the data is public
 ly available was tested and the results obtained generally show low absolu
 te error rates (<5%) between the predicted and reported test accuracies\, 
 which indicate that this method has potential applications for use in mach
 ine learning.\n 
LOCATION:Zoom Meeting ID: 983 5633 9540 Passcode: 268115
END:VEVENT
END:VCALENDAR