kli-casia / ulnno Goto Github PK

#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 on ImageNet fall2011.

The CNN model is GoogeLeNet with [Caffe] (http://caffe.berkeleyvision.org/) implementation. The image format convertor (image2hdf5) is from Toronto Deep Learning.

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

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

• caffe_outputs.py

• cov.py
• whitening.py
• ica.py

• similarity_mat.py
• mds_distance_mat.m

• imagenet_1k_21k_idx.py
• imagenet_zero_shot_unseen_wnids.py
• make_zero_shot_mat.m
• zero_shot_random.py
• zero_shot_pca.py
• zero_shot_ica.py

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