Options

Mining Hard negatives in a Learner possible?

Fred12Fred12 Member Posts: 344 Unicorn
edited November 2018 in Help

hi, 

I read this article about face recognition:

http://cs.brown.edu/courses/cs143/results/proj4/valayshah/

there was a paragraph about training a SVM to mine hard negatives, which improved recognition results of faces a little bit...

my question: is this also possible in rapidminer? how can one train a learner for hard negatives in general? usually, I want to classify the true positives, how can I tell a learner, that this is a "hard negative"?

 

and is this possible not for image data only, but for example sets (tables?). Like for a dataset with parameters in columns... excel sheet or so?

Answers

  • Options
    Thomas_OttThomas_Ott RapidMiner Certified Analyst, RapidMiner Certified Expert, Member Posts: 1,761 Unicorn

    You can use the Threshold operators to set/create a threshold for "crisp" classification. Check them out, they're in the Scoring folder.

Sign In or Register to comment.