Learning To Predict rare events in event sequences. By Gary M Weiss and Haym Hirsh Presented by Veena Raja Rathna. Contents. Aim & Introduction to the problem Basic problem Formulation Definitions & Evaluation metrics Learning methods
Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author.While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server.
Gary M Weiss and Haym Hirsh
Veena Raja Rathna
they require numerical features
do not support predicting a specific ‘event’ within window of time.
It is the percentage of target events correctly predicted
TP - true predictions FP - false Predictions
Simple precision is the percentage of predictions that are correct
Replaces # of correct predictions with target
events correctly predicted
fitness = ((* )+1)precision recall
count j=(1 -distance(j k))3
The similarity of 2 individuals is measured using a phenotypic distance metric that measures the distance based on the performance of the individual
Format: Integer valued timestamp,colon,comma separated list of feature values.2 feature values per event and each event can take on the values a,b,c or d.The first can also take on the value “crash”. The target event is any event with ‘crash’ as the first feature. Warning time is 2 secs and monitoring time is 8 secs.
TW-GA training on this might produce PP: 4:|c,c|*|c,c|.