Skip to content
/ ULNNO Public
forked from yaolubrain/ULNNO

Unsupervised Learning on Neural Network Outputs

Notifications You must be signed in to change notification settings

wangg12/ULNNO

Repository files navigation

#Unsupervised Learning on Neural Network Outputs This repo contains the experiment code in paper

Unsupervised Learning on Neural Network Outputs

The paper presents a new zero-shot learning method, which achieves the state-of-the-art results in ImageNet fall2011.

Instructions

Download the following files from http://image-net.org/

  • ILSVRC2012_img_train.tar (138G)
  • ILSVRC2012_img_val.tar (6.3G)
  • fall11_whole.tar (1.2T)

prepare the images intro HDF5 files, use

  • uncompress.sh
  • correct_format.sh
  • image2hdf5.sh

compute the CNN outputs of GoogLeNet of the images, use

  • caffe_outputs.py

compute PCA and ICA on the CNN outputs, use

  • cov.py
  • whitening.py
  • ica.py

compute the MDS features of WordNet graph, use

  • similarity_mat.py
  • mds_distance_mat.m

run zero-shot learning experiments, use

  • make_zero_shot_mat.m
  • zero_shot_random.py
  • zero_shot_pca.py
  • zero_shot_ica.py

Questions

If you have any question regarding the code and the experiments, please contact me ([email protected]). I would like to hear from you!

About

Unsupervised Learning on Neural Network Outputs

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 87.0%
  • MATLAB 8.2%
  • Shell 4.0%
  • M 0.8%