update paper link

Author: kevinlin311tw

README.md

@@ -11,7 +11,8 @@ This paper presents a simple yet effective supervised deep hash approach that co
 
 <img src="https://www.csie.ntu.edu.tw/~r01944012/ssdh_intro.png" width="800">
 
-The details can be found in the following [arXiv preprint.](http://arxiv.org/abs/1507.00101)
+The TPAMI pre-print can be found in the following [pdf.](https://docs.google.com/viewer?a=v&pid=sites&srcid=ZGVmYXVsdGRvbWFpbnxrZXZpbmxpbjMxMXR3fGd4OjUxMmUwZjAzMjA0YmYyMTg)
+Previous arXiv pre-print can be found in the following [arXiv preprint.](http://arxiv.org/abs/1507.00101)
 Presentation slide can be found [here](http://www.csie.ntu.edu.tw/~r01944012/deepworkshop-slide.pdf)
 
 
@@ -124,7 +125,7 @@ It should be easy to train the model using another dataset as long as that datas
 
 If `./prepare.sh` fails to download data, you may manually download the resouces from:
 
-0. 48-bit SSDH model: [MEGA](https://mega.nz/#!9JMBlCaS!zsTl7eZRMdi25gkLWpj_Uv8LfN_2gQ-UF8OBMhio_3s), [DropBox](https://www.dropbox.com/s/6iqyz1mdhadhzbu/SSDH48_iter_50000.caffemodel?dl=0), [BaiduYun coming soon]
+0. 48-bit SSDH model: [MEGA](https://mega.nz/#!9JMBlCaS!zsTl7eZRMdi25gkLWpj_Uv8LfN_2gQ-UF8OBMhio_3s), [DropBox](https://www.dropbox.com/s/6iqyz1mdhadhzbu/SSDH48_iter_50000.caffemodel?dl=0)
 
 0. CIFAR10 dataset (jpg format): [MEGA](https://mega.nz/#!RENV1bhZ!x0uFnAkqUSTJzKr6HzeeNV9mtDjlgQ0x6ZaXfpxbJkw), [DropBox](https://www.dropbox.com/s/f7q3bbgvat2q1u2/cifar10-dataset.zip?dl=0), [BaiduYun](http://pan.baidu.com/s/1pKsSK7h)