zslearning icon indicating copy to clipboard operation
zslearning copied to clipboard

Error in running main.m

Open yutingchen1211 opened this issue 10 years ago • 2 comments

Hi Milind and Richard,

This is Yuting, a phd student from Boston University. We are trying to compare our recent work with your 13'NIPS paper on zero-shot learning. We try to run the code you provided, however it seems to be some errors in running the main script:

  1. Line 88, there doesn't seem to be a "resolution" field for variable "fullParams". I set it to 11 as for the Gaussian classifier. Is that OK?
  2. Line 152, the variables "thetaSeenSoftmax", "thetaUnseenSoftmax", "thetaMapping", "seenSmTrainParams", "unseenSmTrainParams","mapTrainParams", "validX" and "validY" do not seem to exist. I see from line 44 and line 53 that you saved some variables to "thetaSeenSoftmax.mat" and "thetaUnseenSoftmax.mat", but still there's no variable named after these mat file names... As for the other variables, I did not find any trace of them... Did I miss anything?

Thanks a lot for your code and time. We look forward to hearing back from you :)

Yuting

yutingchen1211 avatar Jan 20 '15 01:01 yutingchen1211

Have u figured it out? if so , I would appreciate it if u could please email me.( [email protected])

mobyIsMe avatar Dec 11 '15 08:12 mobyIsMe

I changed the code on line 152 to the following based on my understanding:

[ guessedCategories, results ] = evaluateGaussianBayesian(thetaSeen, thetaUnseen, ... theta, trainParamsSeen, trainParamsUnseen, trainParams, testX, ... testY, cutoffs, zeroCategories, nonZeroCategories, label_names, wordTable, true);

But it gave me the following error: error: randi: require IMAX >= 1

tengerye avatar Feb 22 '16 14:02 tengerye