I have a NN I'd like to train using supervised learning. Some samples of the training set, however, have better "quality" than others, so I'd like the algorithm to pay "special attention" to them. As general question, how to take this into account in implementation?

Being more specific, I'm working with OpenCV and noticed that the train method apparently have such parameter:

cv2.ANN_MLP.train(inputs, outputs, sampleWeights[, sampleIdx[, params[, flags]]]) → retval


sampleWeights – (RPROP only) Optional floating-point vector of weights for each sample. Some samples may be more important than others for training. You may want to raise the weight of certain classes to find the right balance between hit-rate and false-alarm rate, and so on.

However OpenCV documentation is unclear on this, so how to handle this parameter?



You must log in to answer this question.

Browse other questions tagged .