Hi Mark, I update and delete the redundant files. Thanks for you advice. I've seen the nupic.research code for MNIST, it use net framework but the code level is too abstract for me. I would like to design a hardware architecture so I want to do experiment in more detailed level. BTW, PR means precision-recall?
Thanks, Gordon 2015-03-26 17:39 GMT+08:00 Marek Otahal <[email protected]>: > Hi Gordon, > > nice results! > Can I have a couple of questions? Did use see/use the MNIST code from > nupic.research? > Do you want to make a PR of your results to nupic-community/nupic-vision? > And, can you remove (and add to .gitignore) the files *.swp *.swo *.pyc - > they are useless > > Cheers, > Mark > > On Thu, Mar 26, 2015 at 10:34 AM, 曾國維 <[email protected]> wrote: > >> Hi everyone, >> I use NUPIC spatial pooler for image recognition. >> Here's my code test on MNIST: got about 92% acc. >> https://github.com/tsenggordon/nupic.recogn >> >> I want to extend the code to CIFAR 10 dataset. >> I filter the input by max-pooling the pixel value. >> Ex: [255 123 24 22] -> [1 0 0 0] >> The code is similar to MNIST one >> When I am training the net, I got 99% acc. >> However, while I test the net using test set, I got 16% acc. (very poor >> result) >> >> Any comments are appreciated. >> >> Thanks, >> >> Gordon >> > > > > -- > Marek Otahal :o) >
