- Email: [email protected]
Large margin learning of hierarchical semantic similarity for image classification
Accepted Manuscript Large margin learning of hierarchical semantic similarity for image classification Ju Yong Chang, Kyoung Mu Lee PII: DOI: Reference:
S1077-3142(14)00226-4 http://dx.doi.org/10.1016/j.cviu.2014.11.006 YCVIU 2195
To appear in:
Computer Vision and Image Understanding
Received Date: Accepted Date:
25 March 2013 27 November 2014
Please cite this article as: J.Y. Chang, K.M. Lee, Large margin learning of hierarchical semantic similarity for image classification, Computer Vision and Image Understanding (2014), doi: http://dx.doi.org/10.1016/j.cviu.2014.11.006
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
accuracy along batches 0.45
0.4
0.35
accuracy
0.3
0.25
0.2
0.15
0.1
Ridge Taxem SVM SVMtax HPS Ours
0.05
0
1
2
3 dataset number
4
5
hier cost along batches 9
8
7
hierarchical cost
6
5
4
3
2
Ridge Taxem SVM SVMtax HPS Ours
1
0
1
2
3 dataset number
4
5
precision along batches 0.7
0.6
precision
0.5
0.4
0.3
0.2 Ridge Taxem SVM SVMtax HPS Ours
0.1
0
1
2
3 dataset number
4
5
accuracy along batches 0.4
0.35
0.3
accuracy
0.25
0.2
0.15
0.1 Ridge Taxem SVM SVMtax HPS Ours
0.05
0
1
2
3 dataset number
4
5
hier cost along batches 9
8
7
hierarchical cost
6
5
4
3
2
Ridge Taxem SVM SVMtax HPS Ours
1
0
1
2
3 dataset number
4
5
precision along batches 0.7
0.6
precision
0.5
0.4
0.3
0.2 Ridge Taxem SVM SVMtax HPS Ours
0.1
0
1
2
3 dataset number
4
5
accuracy along batches 0.14
0.12
accuracy
0.1
0.08
0.06
0.04
0.02 SVM HPS Ours 0
1
2 dataset number
hier cost along batches 12
10
hierarchical cost
8
6
4
2 SVM HPS Ours 0
1
2 dataset number precision along batches
0.25
0.2
precision
0.15
0.1
0.05 SVM HPS Ours 0
1
2 dataset number
accuracy along labels 1
0.9
0.8
0.7
accuracy
0.6
0.5
0.4
SVM HPS Ours
0.3
0.2
0.1
0
0
100
200
300
400 500 600 number of labels
700
800
900
1000
hier cost along labels 12
10
hierarchical cost
8
SVM HPS Ours
6
4
2
0
0
100
200
300
400 500 600 number of labels
700
800
900
1000
> Novel large margin formulation for semantic similarity learning. > Efficient optimization algorithm to solve the proposed semi-definite program (SDP). > Thorough experimental study to compare the performances of several algorithms for hierarchical image classification. > State-of-the-art classification performance under the hierarchical-loss criterion.
S1077-3142(14)00226-4 http://dx.doi.org/10.1016/j.cviu.2014.11.006 YCVIU 2195
To appear in:
Computer Vision and Image Understanding
Received Date: Accepted Date:
25 March 2013 27 November 2014
Please cite this article as: J.Y. Chang, K.M. Lee, Large margin learning of hierarchical semantic similarity for image classification, Computer Vision and Image Understanding (2014), doi: http://dx.doi.org/10.1016/j.cviu.2014.11.006
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
accuracy along batches 0.45
0.4
0.35
accuracy
0.3
0.25
0.2
0.15
0.1
Ridge Taxem SVM SVMtax HPS Ours
0.05
0
1
2
3 dataset number
4
5
hier cost along batches 9
8
7
hierarchical cost
6
5
4
3
2
Ridge Taxem SVM SVMtax HPS Ours
1
0
1
2
3 dataset number
4
5
precision along batches 0.7
0.6
precision
0.5
0.4
0.3
0.2 Ridge Taxem SVM SVMtax HPS Ours
0.1
0
1
2
3 dataset number
4
5
accuracy along batches 0.4
0.35
0.3
accuracy
0.25
0.2
0.15
0.1 Ridge Taxem SVM SVMtax HPS Ours
0.05
0
1
2
3 dataset number
4
5
hier cost along batches 9
8
7
hierarchical cost
6
5
4
3
2
Ridge Taxem SVM SVMtax HPS Ours
1
0
1
2
3 dataset number
4
5
precision along batches 0.7
0.6
precision
0.5
0.4
0.3
0.2 Ridge Taxem SVM SVMtax HPS Ours
0.1
0
1
2
3 dataset number
4
5
accuracy along batches 0.14
0.12
accuracy
0.1
0.08
0.06
0.04
0.02 SVM HPS Ours 0
1
2 dataset number
hier cost along batches 12
10
hierarchical cost
8
6
4
2 SVM HPS Ours 0
1
2 dataset number precision along batches
0.25
0.2
precision
0.15
0.1
0.05 SVM HPS Ours 0
1
2 dataset number
accuracy along labels 1
0.9
0.8
0.7
accuracy
0.6
0.5
0.4
SVM HPS Ours
0.3
0.2
0.1
0
0
100
200
300
400 500 600 number of labels
700
800
900
1000
hier cost along labels 12
10
hierarchical cost
8
SVM HPS Ours
6
4
2
0
0
100
200
300
400 500 600 number of labels
700
800
900
1000
> Novel large margin formulation for semantic similarity learning. > Efficient optimization algorithm to solve the proposed semi-definite program (SDP). > Thorough experimental study to compare the performances of several algorithms for hierarchical image classification. > State-of-the-art classification performance under the hierarchical-loss criterion.