-
Notifications
You must be signed in to change notification settings - Fork 2
/
CBM.html
132 lines (130 loc) · 5.39 KB
/
CBM.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
<!DOCTYPE html>
<html>
<head>
<meta charset="utf-8">
<title>CBM</title>
<link rel="stylesheet" type="text/css" href="assets/scripts/bulma.min.css">
<link rel="stylesheet" type="text/css" href="assets/scripts/theme.css">
<link rel="stylesheet" type="text/css" href="https://cdn.bootcdn.net/ajax/libs/font-awesome/4.7.0/css/font-awesome.min.css">
</head>
<body>
<section class="hero is-light" style="">
<div class="hero-body" style="padding-top: 50px;">
<div class="container" style="text-align: center;margin-bottom:5px;">
<h1 class="title">
Cooperative Bi-path Metric for Few-shot Learning
</h1>
<div class="author">Zeyuan Wang<sup>1</sup></div>
<div class="author">Yifan Zhao<sup>1</sup></div>
<div class="author">Jia Li<sup>1,3,4</sup>*</div>
<div class="author">Yonghong Tian<sup>2,4</sup></div>
<div class="group">
<a href="http://cvteam.net/">CVTEAM</a>
</div>
<div class="aff">
<p><sup>1</sup>State Key Laboratory of Virtual Reality Technology and Systems, SCSE, Beihang University, Beijing, 100191, China</p>
<p><sup>2</sup>School of Electronics Engineering and Computer Science, Peking University, Beijing, 100871, China</p>
<p><sup>3</sup>Beijing Advanced Innovation Center for Big Data and Brain Computing, Beihang University, Beijing, 100191, China</p>
<p><sup>4</sup>Peng Cheng Laboratory, Shenzhen, 518066, China</p>
</div>
<div class="con">
<p style="font-size: 24px; margin-top:5px; margin-bottom: 15px;">
ACM MM 2020
</p>
</div>
<div class="columns">
<div class="column"></div>
<div class="column"></div>
<div class="column">
<a href="https://arxiv.org/abs/2008.04031" target="_blank">
<p class="link">Paper</p>
</a>
</div>
<div class="column">
<p class="link">Code</p>
</div>
<div class="column"></div>
<div class="column"></div>
</div>
</div>
</div>
</section>
<div style="text-align: center;">
<div class="container" style="max-width:850px">
<div style="text-align: center;">
<img src="assets/CBM/approach.jpg" class="centerImage">
</div>
</div>
<div class="head_cap">
<p style="color:gray;">
The approach of CBM
</p>
</div>
</div>
<section class="hero">
<div class="hero-body">
<div class="container" style="max-width: 800px" >
<h1 style="">Abstract</h1>
<p style="text-align: justify; font-size: 17px;">
Given base classes with sufficient labeled samples, the target of
few-shot classification is to recognize unlabeled samples of novel
classes with only a few labeled samples. Most existing methods only
pay attention to the relationship between labeled and unlabeled
samples of novel classes, which do not make full use of information
within base classes. In this paper, we make two contributions to
investigate the few-shot classification problem. First, we report a
simple and effective baseline trained on base classes in the way
of traditional supervised learning, which can achieve comparable
results to the state of the art. Second, based on the baseline, we propose a cooperative bi-path metric for classification, which leverages
the correlations between base classes and novel classes to further
improve the accuracy. Experiments on two widely used benchmarks
show that our method is a simple and effective framework, and
a new state of the art is established in the few-shot classification
field.
</p>
</div>
</div>
</section>
<section class="hero is-light" style="background-color:#FFFFFF;">
<div class="hero-body">
<div class="container" style="max-width:800px;margin-bottom:20px;">
<h1>
Experiment
</h1>
</div>
<div class="container" style="max-width:800px">
<div style="text-align: center;">
<img src="assets/CBM/experiment.jpg" class="centerImage">
</div>
</div>
</div>
</section>
<section class="hero" style="padding-top:0px;">
<div class="hero-body">
<div class="container" style="max-width:800px;">
<div class="card">
<header class="card-header">
<p class="card-header-title">
BibTex Citation
</p>
<a class="card-header-icon button-clipboard" style="border:0px; background: inherit;" data-clipboard-target="#bibtex-info" >
<i class="fa fa-copy" height="20px"></i>
</a>
</header>
<div class="card-content">
<pre style="background-color:inherit;padding: 0px;" id="bibtex-info">@misc{wang2020cooperative,
title={Cooperative Bi-path Metric for Few-shot Learning},
author={Zeyuan Wang and Yifan Zhao and Jia Li and Yonghong Tian},
year={2020},
eprint={2008.04031},
archivePrefix={arXiv},
primaryClass={cs.CV}
}</pre>
</div>
</section>
<script type="text/javascript" src="assets/scripts/clipboard.min.js"></script>
<script>
new ClipboardJS('.button-clipboard');
</script>
</body>
</html>