-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathindex.html
209 lines (184 loc) · 7.89 KB
/
index.html
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
<!DOCTYPE html>
<html>
<head>
<meta charset="UTF-8">
<meta name="viewport" content="width=device-width, maximum-scale=1.0, initial-scale=1.0">
<title>Can We Find Strong Lottery Tickets in Generative Models?</title>
<link rel="stylesheet" href="index.css">
<meta property="og:image" content="Path to my teaser.png"/> <!-- Facebook automatically scrapes this. Go to https://developers.facebook.com/tools/debug/ if you update and want to force Facebook to rescrape. -->
<meta property="og:title" content="Can We Find Strong Lottery Tickets in Generative Models?" />
<meta property="og:description" content="Paper description." />
<!-- Get from Google Analytics -->
<!-- Global site tag (gtag.js) - Google Analytics -->
<script src="http://www.google.com/jsapi" type="text/javascript"></script>
<script type="text/javascript">google.load("jquery", "1.3.2");</script>
<script async src=""></script>
<script src="main.js" ></script>
<script>
window.dataLayer = window.dataLayer || [];
function gtag(){dataLayer.push(arguments);}
gtag('js', new Date());
gtag('config', 'UA-75863369-6');
</script>
</head>
<body>
<div id="wrap">
<section>
<center>
<div class="wrapper">
<span style="font-size:36px;"> Can We Find Strong Lottery Tickets in Generative Models? (AAAI'23)</span>
<div class="container" style="margin-top: 2em;" >
<div class="item">
<span style="font-size:20px"><a href="https://scholar.google.com/citations?hl=ko&user=O5wO-uYAAAAJ">Sangyeop Yeo<sup>1</sup></a></span>
</div>
<div class="item">
<span style="font-size:20px">Yoojin Jang<sup>1</sup></a></span>
</div>
<div class="item">
<span style="font-size:20px"><a href="https://scholar.google.co.kr/citations?hl=en&user=Cs75s1MAAAAJ&view_op=list_works&sortby=pubdate">Jy-yong Sohn<sup>2</sup></a></span>
</div>
<div class="item">
<span style="font-size:20px"><a href="https://scholar.google.com/citations?user=jcP7m1QAAAAJ&hl=en">Dongyoon Han<sup>3</sup></a></a></span>
</div>
<div class="item">
<span style="font-size:20px"><a href="https://scholar.google.co.kr/citations?hl=en&user=7NBlQw4AAAAJ">Jaejun Yoo<sup>1</sup></a></span>
</div>
</div>
<div class="container desc">
<div class="item">
<span style="font-size:15px"><sup>1</sup>Ulsan National Institute of Science & Technology</span>
</div>
<div class="item">
<span style="font-size:15px"><sup>2</sup>University of Wisconsin-Madison</span>
</div>
<div class="item">
<span style="font-size:15px"><sup>3</sup>NAVER AI Lab</span>
</div>
</div>
<div class="container">
<div class="item">
<span style="font-size:24px"><a href="https://arxiv.org/abs/2212.08311" target="_blank">[Paper]</a></span>
</div>
<div class="item">
<span style="font-size:24px"><a href="https://youtu.be/8bvk0LzoVB4" target="_blank">[Video]</a></span>
</div>
<div class="item">
<span style="font-size:24px"><a href="https://drive.google.com/file/d/1DToxSE_wQghAt8XVGITncVBSe8dyyE3K/view?usp=share_link" target="_blank">[Poster]</a></span>
</div>
<div class="item">
<span style="font-size:24px"><a href="https://github.com/SangyeopYeo/Edge-popup_with_MMD">[Github]</a></span>
</div>
<div class="container">
</div>
</div>
</div>
<div class="overview">
<img class="round" style="width:90%;" src="./resources/figure2.jpg"/>
</div>
<div class="wrapper abs">
<hr>
<table align=center width=100%;>
<center><h1>Abstract</h1></center>
<tr>
<td>
<b>Yes.</b> In this paper, we investigate <i>strong lottery tickets</i> in generative models, the subnetworks that achieve good generative performance
without any weight update. Neural network pruning is considered the main cornerstone of model compression for reducing the costs of computation and memory.
Unfortunately, pruning a generative model has not been extensively explored, and all existing pruning algorithms suffer from excessive weight-training costs, performance degradation, limited generalizability, or complicated training. To address these problems, we propose to find a strong lottery ticket via moment-matching scores. Our experimental results show that the discovered subnetwork can perform similarly or better than the trained dense model even when only 10% of the weights remain. To the best of our knowledge, we are the first to show the existence of strong lottery tickets in generative models and provide an algorithm to find it stably.
</td>
</tr>
</table>
</div>
<br>
<hr>
<center><h1>Results</h1></center>
<div class="results">
<table align="center" width=100%; >
<td colspan="2"> <p style="word-wrap: break-word;"> The sparsity
<b>"0.k"</b>
means the ratio of remaining weights so if the sparsity is higher, the ratio of remaining weights is higher.</p></td>
<tr ></tr>
<td colspan="2"><p align="center">GFMN - LSUN Bedroom Dataset</p></td>
<tr >
<td>
<img src="resources/gfmn/sparsity0.1_gfmn.jpg" width="100%;">
</td>
<td>
<img src="resources/gfmn/fully_trained_gfmn.jpg" width="100%;">
</td>
</tr>
<td colspan="2"><p align="center">BigGAN - CelebA Dataset</p></td>
<tr >
<td>
<img src="resources/biggan/sparsity0.1_biggan.jpg" width="100%">
</td>
<td>
<img src="resources/biggan/fully_trained_biggan.jpg" width="100%">
</td>
</tr>
<td colspan="2"><p align="center">SNGAN - CelebA Dataset</p></td>
<tr>
<td>
<img src="resources/sngan/sparsity0.1_sngan.jpg" width="100%">
</td>
<td>
<img src="resources/sngan/fully_trained_sngan.jpg" width="100%">
</td>
</tr>
</table>
</div>
<div class="wrapper supple" style="padding-top: 1em;">
<hr>
<table align=center width=90% >
<center><h1>Paper and Supplementary Material</h1></center>
<tr>
<td><a href="https://arxiv.org/abs/2212.08311" target="_blank"><img class="layered-paper-big supple_img" style="height:175px" src="./resources/paper.png"/></a></td>
<td><span >Sangyeop Yeo, Yoojin Jang, Jy-yong Sohn, Dongyoon Han, Jaejun Yoo.<br>
<b>Can We Find Strong Lottery Tickets in Generative Models?</b><br>
AAAI 2023 Accepted<br>
Arxiv
[<a href="https://arxiv.org/abs/2212.08311" target="_blank">Link</a>] <!-- (<a href="./resources/camera-ready.pdf">camera ready</a>)<br> -->
<span style="font-size:4pt"><a href=""><br></a>
</span>
</td>
</tr>
</table>
<!--<table align=center width=600px>
<tr>
<td><span style="font-size:14pt"><center>
<a href="./resources/bibtex.txt">[Bibtex]</a>
</center></td>
</tr>
</table>
-->
<br>
<br>
<hr>
<table align=center width=90%>
<tr>
<td width=400px>
<left>
<center><h1>Acknowledgements</h1></center>
This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT) (No. 2.220574.01),
Institute of Information & communications Technology Planning &
Evaluation (IITP) grant funded by the Korea government (MSIT)
(No.2020-0-01336, Artificial Intelligence Graduate School Program (UNIST)),
and Institue of Information & communications Technology Planning &
Evaluation (IITP) grant funded by the Korea government (MSIT) (No.2022-0-00959,
(Part 2) Few-Shot Learning of Causal Inference in Vision and Language for Decision Making).
And This template was originally made by <a href="http://web.mit.edu/phillipi/">Phillip Isola</a> and <a href="http://richzhang.github.io/">Richard Zhang</a> for a <a href="http://richzhang.github.io/colorization/">colorful</a> ECCV project; the code can be found <a href="https://github.com/richzhang/webpage-template">here</a>.
</left>
</td>
</tr>
</table>
</div>
</section>
<footer>
<center>
<p style="color: white; ">Lab. of Advanced Imaging Tech</p>
<a style="width: fit-content" href="https://sites.google.com/view/jaejunyoo"><img style="display:block; margin: 0;"
src="resources/others/logo_lait.png" width="50px"></a>
</center>
</footer>
</div>
</body>
</html>