polar: attention-based cnn for one-shot personalized ... · 1vinyals et al., matching networks for...
TRANSCRIPT
![Page 1: POLAR: Attention-based CNN for One-shot Personalized ... · 1Vinyals et al., Matching Networks for One Shot Learning. 6/20. Introduction & Problem De nitionApproachExperiments & Analysis](https://reader034.vdocuments.site/reader034/viewer/2022050409/5f85e49ba1d3a8189b46dba2/html5/thumbnails/1.jpg)
Introduction & Problem Definition Approach Experiments & Analysis
POLAR: Attention-based CNN for One-shotPersonalized Article Recommendation
Zhengxiao Du, Jie Tang, Yuhui Ding
Tsinghua University
{duzx16, dingyh15}@mails.tsinghua.edu.cn,[email protected]
September 13, 2018
1 / 20
![Page 2: POLAR: Attention-based CNN for One-shot Personalized ... · 1Vinyals et al., Matching Networks for One Shot Learning. 6/20. Introduction & Problem De nitionApproachExperiments & Analysis](https://reader034.vdocuments.site/reader034/viewer/2022050409/5f85e49ba1d3a8189b46dba2/html5/thumbnails/2.jpg)
Introduction & Problem Definition Approach Experiments & Analysis
Motivation
The publication output is growing every year(data source: DBLP)
1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 20160k
50k
100k
150k
200k
250k
300kBook and ThesesConference and Workshop PapersEditorshipInformal PublicationsJournal ArticlesParts in Books or CollectionsReference Works
2 / 20
![Page 3: POLAR: Attention-based CNN for One-shot Personalized ... · 1Vinyals et al., Matching Networks for One Shot Learning. 6/20. Introduction & Problem De nitionApproachExperiments & Analysis](https://reader034.vdocuments.site/reader034/viewer/2022050409/5f85e49ba1d3a8189b46dba2/html5/thumbnails/3.jpg)
Introduction & Problem Definition Approach Experiments & Analysis
Related-Article Recommendation
Figure: An example from AMiner.org
3 / 20
![Page 4: POLAR: Attention-based CNN for One-shot Personalized ... · 1Vinyals et al., Matching Networks for One Shot Learning. 6/20. Introduction & Problem De nitionApproachExperiments & Analysis](https://reader034.vdocuments.site/reader034/viewer/2022050409/5f85e49ba1d3a8189b46dba2/html5/thumbnails/4.jpg)
Introduction & Problem Definition Approach Experiments & Analysis
Challenge
How to provide personalized and non-personalizedrecommendation?
How to overcome the sparsity of user feedback?
How to utilize representative texts of articles effectively?
4 / 20
![Page 5: POLAR: Attention-based CNN for One-shot Personalized ... · 1Vinyals et al., Matching Networks for One Shot Learning. 6/20. Introduction & Problem De nitionApproachExperiments & Analysis](https://reader034.vdocuments.site/reader034/viewer/2022050409/5f85e49ba1d3a8189b46dba2/html5/thumbnails/5.jpg)
Introduction & Problem Definition Approach Experiments & Analysis
Problem Definition
Definition
One-shot Personalized Article Recommendation Problem
Input: query article dqcandidate set D = {d1, d2, · · · , dN}support set S = {(di , yi )}Ti=1 related to user u
Output: a totally ordered set R(dq,S) ⊂ D with |R| = k
5 / 20
![Page 6: POLAR: Attention-based CNN for One-shot Personalized ... · 1Vinyals et al., Matching Networks for One Shot Learning. 6/20. Introduction & Problem De nitionApproachExperiments & Analysis](https://reader034.vdocuments.site/reader034/viewer/2022050409/5f85e49ba1d3a8189b46dba2/html5/thumbnails/6.jpg)
Introduction & Problem Definition Approach Experiments & Analysis
Problem Definition
Definition
One-shot Personalized Article Recommendation Problem
Input: query article dqcandidate set D = {d1, d2, · · · , dN}support set S = {(di , yi )}Ti=1 related to user u
Output: a totally ordered set R(dq,S) ⊂ D with |R| = k
5 / 20
![Page 7: POLAR: Attention-based CNN for One-shot Personalized ... · 1Vinyals et al., Matching Networks for One Shot Learning. 6/20. Introduction & Problem De nitionApproachExperiments & Analysis](https://reader034.vdocuments.site/reader034/viewer/2022050409/5f85e49ba1d3a8189b46dba2/html5/thumbnails/7.jpg)
Introduction & Problem Definition Approach Experiments & Analysis
Problem Definition
Definition
One-shot Personalized Article Recommendation Problem
Input: query article dqcandidate set D = {d1, d2, · · · , dN}support set S = {(di , yi )}Ti=1 related to user u
Output: a totally ordered set R(dq, S) ⊂ D with |R| = k
5 / 20
![Page 8: POLAR: Attention-based CNN for One-shot Personalized ... · 1Vinyals et al., Matching Networks for One Shot Learning. 6/20. Introduction & Problem De nitionApproachExperiments & Analysis](https://reader034.vdocuments.site/reader034/viewer/2022050409/5f85e49ba1d3a8189b46dba2/html5/thumbnails/8.jpg)
Introduction & Problem Definition Approach Experiments & Analysis
One-shot Learning
Image Classification1
y =k∑
i=1
a(x , xi )yi
Article Recommendation
Query article dq
Support set{(di , yi )}Ti=1
si = c(dq, di ) + 1T
∑Tj=1 c(di , dj)yj
the matching to the queryarticle
the matching to the userpreference(maybe missing)
1Vinyals et al., Matching Networks for One Shot Learning.6 / 20
![Page 9: POLAR: Attention-based CNN for One-shot Personalized ... · 1Vinyals et al., Matching Networks for One Shot Learning. 6/20. Introduction & Problem De nitionApproachExperiments & Analysis](https://reader034.vdocuments.site/reader034/viewer/2022050409/5f85e49ba1d3a8189b46dba2/html5/thumbnails/9.jpg)
Introduction & Problem Definition Approach Experiments & Analysis
One-shot Learning
Image Classification1
y =k∑
i=1
a(x , xi )yi
Article Recommendation
Query article dq
Support set{(di , yi )}Ti=1
si = c(dq, di ) +
1T
∑Tj=1 c(di , dj)yj
the matching to the queryarticle
the matching to the userpreference(maybe missing)
1Vinyals et al., Matching Networks for One Shot Learning.6 / 20
![Page 10: POLAR: Attention-based CNN for One-shot Personalized ... · 1Vinyals et al., Matching Networks for One Shot Learning. 6/20. Introduction & Problem De nitionApproachExperiments & Analysis](https://reader034.vdocuments.site/reader034/viewer/2022050409/5f85e49ba1d3a8189b46dba2/html5/thumbnails/10.jpg)
Introduction & Problem Definition Approach Experiments & Analysis
One-shot Learning
Image Classification1
y =k∑
i=1
a(x , xi )yi
Article Recommendation
Query article dq
Support set{(di , yi )}Ti=1
si =
c(dq, di ) +
1T
∑Tj=1 c(di , dj)yj
the matching to the queryarticle
the matching to the userpreference(maybe missing)
1Vinyals et al., Matching Networks for One Shot Learning.6 / 20
![Page 11: POLAR: Attention-based CNN for One-shot Personalized ... · 1Vinyals et al., Matching Networks for One Shot Learning. 6/20. Introduction & Problem De nitionApproachExperiments & Analysis](https://reader034.vdocuments.site/reader034/viewer/2022050409/5f85e49ba1d3a8189b46dba2/html5/thumbnails/11.jpg)
Introduction & Problem Definition Approach Experiments & Analysis
One-shot Learning
Image Classification1
y =k∑
i=1
a(x , xi )yi
Article Recommendation
Query article dq
Support set{(di , yi )}Ti=1
si = c(dq, di ) + 1T
∑Tj=1 c(di , dj)yj
the matching to the queryarticle
the matching to the userpreference(maybe missing)
1Vinyals et al., Matching Networks for One Shot Learning.6 / 20
![Page 12: POLAR: Attention-based CNN for One-shot Personalized ... · 1Vinyals et al., Matching Networks for One Shot Learning. 6/20. Introduction & Problem De nitionApproachExperiments & Analysis](https://reader034.vdocuments.site/reader034/viewer/2022050409/5f85e49ba1d3a8189b46dba2/html5/thumbnails/12.jpg)
Introduction & Problem Definition Approach Experiments & Analysis
Architecture
EmbeddingLayer
Candidate
kwdld
kwd2
kwd1
· · ·
Query
kwqlq
kwq2
kwq1
· · ·
PersonalizedScore
7 / 20
![Page 13: POLAR: Attention-based CNN for One-shot Personalized ... · 1Vinyals et al., Matching Networks for One Shot Learning. 6/20. Introduction & Problem De nitionApproachExperiments & Analysis](https://reader034.vdocuments.site/reader034/viewer/2022050409/5f85e49ba1d3a8189b46dba2/html5/thumbnails/13.jpg)
Introduction & Problem Definition Approach Experiments & Analysis
Architecture
EmbeddingLayer
Candidate
kwdld
kwd2
kwd1
· · ·
Query
kwqlq
kwq2
kwq1
· · ·
MatchingMatrix
AttentionMatrix
ConvInput
PersonalizedScore
7 / 20
![Page 14: POLAR: Attention-based CNN for One-shot Personalized ... · 1Vinyals et al., Matching Networks for One Shot Learning. 6/20. Introduction & Problem De nitionApproachExperiments & Analysis](https://reader034.vdocuments.site/reader034/viewer/2022050409/5f85e49ba1d3a8189b46dba2/html5/thumbnails/14.jpg)
Introduction & Problem Definition Approach Experiments & Analysis
Architecture
EmbeddingLayer
Candidate
kwdld
kwd2
kwd1
· · ·
Query
kwqlq
kwq2
kwq1
· · ·
MatchingMatrix
AttentionMatrix
FeatureMap
ConvInput
Convolution andMax-Pooling
Hidden State
Full-ConnectedLayer
MatchingScore
PersonalizedScore
7 / 20
![Page 15: POLAR: Attention-based CNN for One-shot Personalized ... · 1Vinyals et al., Matching Networks for One Shot Learning. 6/20. Introduction & Problem De nitionApproachExperiments & Analysis](https://reader034.vdocuments.site/reader034/viewer/2022050409/5f85e49ba1d3a8189b46dba2/html5/thumbnails/15.jpg)
Introduction & Problem Definition Approach Experiments & Analysis
Architecture
EmbeddingLayer
Candidate
kwdld
kwd2
kwd1
· · ·
Query
kwqlq
kwq2
kwq1
· · ·
MatchingMatrix
AttentionMatrix
FeatureMap
ConvInput
Convolution andMax-Pooling
Hidden State
Full-ConnectedLayer
Support Set
d1
y1
d2
y2
dT
yT
MatchingScore
PersonalizedScore
One ShotMatching
7 / 20
![Page 16: POLAR: Attention-based CNN for One-shot Personalized ... · 1Vinyals et al., Matching Networks for One Shot Learning. 6/20. Introduction & Problem De nitionApproachExperiments & Analysis](https://reader034.vdocuments.site/reader034/viewer/2022050409/5f85e49ba1d3a8189b46dba2/html5/thumbnails/16.jpg)
Introduction & Problem Definition Approach Experiments & Analysis
Architecture
EmbeddingLayer
Candidate
kwdld
kwd2
kwd1
· · ·
Query
kwqlq
kwq2
kwq1
· · ·
MatchingMatrix
AttentionMatrix
FeatureMap
ConvInput
Convolution andMax-Pooling
Hidden State
Full-ConnectedLayer
Support Set
d1
y1
d2
y2
dT
yT
MatchingScore
PersonalizedScore
FinalScore
One ShotMatching
7 / 20
![Page 17: POLAR: Attention-based CNN for One-shot Personalized ... · 1Vinyals et al., Matching Networks for One Shot Learning. 6/20. Introduction & Problem De nitionApproachExperiments & Analysis](https://reader034.vdocuments.site/reader034/viewer/2022050409/5f85e49ba1d3a8189b46dba2/html5/thumbnails/17.jpg)
Introduction & Problem Definition Approach Experiments & Analysis
Matching Matrix and Attention Matrix
Matching Matrix:(dm, dn)→ Rlm×ln
the similarity between the words of two articles.
M(m,n)i ,j =
~wTmi · ~wnj
‖~wmi‖ · ‖~wnj‖
Attention Matrix:(dm, dn)→ Rlm×ln
the importance of the matching signals
A(m,n)i ,j = rmi · rnj
8 / 20
![Page 18: POLAR: Attention-based CNN for One-shot Personalized ... · 1Vinyals et al., Matching Networks for One Shot Learning. 6/20. Introduction & Problem De nitionApproachExperiments & Analysis](https://reader034.vdocuments.site/reader034/viewer/2022050409/5f85e49ba1d3a8189b46dba2/html5/thumbnails/18.jpg)
Introduction & Problem Definition Approach Experiments & Analysis
Matching Matrix and Attention Matrix
Matching Matrix:(dm, dn)→ Rlm×ln
the similarity between the words of two articles.
M(m,n)i ,j =
~wTmi · ~wnj
‖~wmi‖ · ‖~wnj‖
Attention Matrix:(dm, dn)→ Rlm×ln
the importance of the matching signals
A(m,n)i ,j = rmi · rnj
8 / 20
![Page 19: POLAR: Attention-based CNN for One-shot Personalized ... · 1Vinyals et al., Matching Networks for One Shot Learning. 6/20. Introduction & Problem De nitionApproachExperiments & Analysis](https://reader034.vdocuments.site/reader034/viewer/2022050409/5f85e49ba1d3a8189b46dba2/html5/thumbnails/19.jpg)
Introduction & Problem Definition Approach Experiments & Analysis
Local Weight and Global Weight
The word weight rt is the product of its local weight and globalweight.
Global Weight: The importance of a word in thecorpus(shared among different articles)
υij = [IDF(tij)]β
The local weight is a little more complicated. . .
9 / 20
![Page 20: POLAR: Attention-based CNN for One-shot Personalized ... · 1Vinyals et al., Matching Networks for One Shot Learning. 6/20. Introduction & Problem De nitionApproachExperiments & Analysis](https://reader034.vdocuments.site/reader034/viewer/2022050409/5f85e49ba1d3a8189b46dba2/html5/thumbnails/20.jpg)
Introduction & Problem Definition Approach Experiments & Analysis
Local Weight and Global Weight
The word weight rt is the product of its local weight and globalweight.
Global Weight: The importance of a word in thecorpus(shared among different articles)
υij = [IDF(tij)]β
The local weight is a little more complicated. . .
9 / 20
![Page 21: POLAR: Attention-based CNN for One-shot Personalized ... · 1Vinyals et al., Matching Networks for One Shot Learning. 6/20. Introduction & Problem De nitionApproachExperiments & Analysis](https://reader034.vdocuments.site/reader034/viewer/2022050409/5f85e49ba1d3a8189b46dba2/html5/thumbnails/21.jpg)
Introduction & Problem Definition Approach Experiments & Analysis
Local Weight and Global Weight
The word weight rt is the product of its local weight and globalweight.
Global Weight: The importance of a word in thecorpus(shared among different articles)
υij = [IDF(tij)]β
The local weight is a little more complicated. . .
9 / 20
![Page 22: POLAR: Attention-based CNN for One-shot Personalized ... · 1Vinyals et al., Matching Networks for One Shot Learning. 6/20. Introduction & Problem De nitionApproachExperiments & Analysis](https://reader034.vdocuments.site/reader034/viewer/2022050409/5f85e49ba1d3a8189b46dba2/html5/thumbnails/22.jpg)
Introduction & Problem Definition Approach Experiments & Analysis
Local Weight
Local Weight: The importance of a word in the article
A neural network is employed to compute the local weight.
The feature vector forword tij
~xij = ~wij − ~w i
1
1
2
2
3
3
4
4
5
5
6
6
7
7
0
The triangular points denote thevectors of the words in two texts
The circular points denote themean vectors of the texts.
The lines with arrowsdenote the feature vectors
10 / 20
![Page 23: POLAR: Attention-based CNN for One-shot Personalized ... · 1Vinyals et al., Matching Networks for One Shot Learning. 6/20. Introduction & Problem De nitionApproachExperiments & Analysis](https://reader034.vdocuments.site/reader034/viewer/2022050409/5f85e49ba1d3a8189b46dba2/html5/thumbnails/23.jpg)
Introduction & Problem Definition Approach Experiments & Analysis
Local Weight Network
The feature vector ~xij represents the semantic differencebetween the article and the term.
Let ~u(L)ij be the output of the last linear layer, the output of
the local weight network is
µij = σ(W (L) · ~u (L)ij + b(L)) + α
α sets a lower bound for local weights.
11 / 20
![Page 24: POLAR: Attention-based CNN for One-shot Personalized ... · 1Vinyals et al., Matching Networks for One Shot Learning. 6/20. Introduction & Problem De nitionApproachExperiments & Analysis](https://reader034.vdocuments.site/reader034/viewer/2022050409/5f85e49ba1d3a8189b46dba2/html5/thumbnails/24.jpg)
Introduction & Problem Definition Approach Experiments & Analysis
CNN & Training
The matching matrix and attention matrix are combined byelement-wise multiplication and sent to a CNN.
MatchingMatrix
AttentionMatrix
FeatureMap
ConvInput
Convolution andMax-Pooling
Hidden State
Full-ConnectedLayer
MatchingScore
The entire model, including the local weight network, istrained on the target task.
12 / 20
![Page 25: POLAR: Attention-based CNN for One-shot Personalized ... · 1Vinyals et al., Matching Networks for One Shot Learning. 6/20. Introduction & Problem De nitionApproachExperiments & Analysis](https://reader034.vdocuments.site/reader034/viewer/2022050409/5f85e49ba1d3a8189b46dba2/html5/thumbnails/25.jpg)
Introduction & Problem Definition Approach Experiments & Analysis
Dataset
AMiner: papers from ArnetMiner1
Patent: patent documents from USPTO
RARD (Related Article Recommendation Dataset2):fromSowiport, a digital library service provider.
1Tang et al. ArnetMiner: Extraction and Mining of Academic SocialNetworks. In SIGKDD’2008.
2Beel et al. Rard: The related-article recommendation dataset (2017)13 / 20
![Page 26: POLAR: Attention-based CNN for One-shot Personalized ... · 1Vinyals et al., Matching Networks for One Shot Learning. 6/20. Introduction & Problem De nitionApproachExperiments & Analysis](https://reader034.vdocuments.site/reader034/viewer/2022050409/5f85e49ba1d3a8189b46dba2/html5/thumbnails/26.jpg)
Introduction & Problem Definition Approach Experiments & Analysis
Experimental Results
Table: Results of recommendation without personalization(%).
AMiner Patent RARDMethod NG@3 NG@5 NG@10 NG@3 NG@5 NG@10 NG@1 NG@3 NG@5
TF-IDF 74.3 81.8 87.5 51.8 56.4 63.4 37.6 39.8 46.3Doc2Vec 60.0 65.8 79.1 44.6 45.6 53.5 28.4 34.0 40.0
WMD 73.0 76.3 86.2 57.4 58.5 61.9 23.4 38.2 46.8
MV-LSTM 56.2 61.2 76.2 60.2 59.0 65.0 22.2 30.7 39.3Duet 66.6 74.4 82.6 54.5 57.5 64.6 22.3 31.1 39.8
DRMM 75.0 79.9 87.1 55.0 56.2 64.7 33.1 36.3 40.6MatchPyramid 73.5 80.0 86.8 56.4 61.4 64.4 29.1 36.2 42.8
POLAR 80.3 85.2 90.1 67.8 69.5 73.6 42.8 46.3 51.5
1For the fairness of comparison, all models don’t involve personalization.2NG stands for NDCG. 14 / 20
![Page 27: POLAR: Attention-based CNN for One-shot Personalized ... · 1Vinyals et al., Matching Networks for One Shot Learning. 6/20. Introduction & Problem De nitionApproachExperiments & Analysis](https://reader034.vdocuments.site/reader034/viewer/2022050409/5f85e49ba1d3a8189b46dba2/html5/thumbnails/27.jpg)
Introduction & Problem Definition Approach Experiments & Analysis
How One-shot Personalization Can Help
Randomly divide the labeled articles into the support set andthe candidate set to recommend.
POLAR-OS is the proposed one-shot framework andPOLAR-ALL the best model that ignores support sets in theprevious part.
Table: Performance for the model with and without personalization.
AMiner Patent RARDMethod NDCG@1 NDCG@3 NDCG@1 NDCG@3 NDCG@1 NDCG@3
POLAR-OS 79.1 81.9 57.1 69.7 39.4 39.2POLAR-ALL 76.1 79.2 52.3 66.2 36.5 36.5
15 / 20
![Page 28: POLAR: Attention-based CNN for One-shot Personalized ... · 1Vinyals et al., Matching Networks for One Shot Learning. 6/20. Introduction & Problem De nitionApproachExperiments & Analysis](https://reader034.vdocuments.site/reader034/viewer/2022050409/5f85e49ba1d3a8189b46dba2/html5/thumbnails/28.jpg)
Introduction & Problem Definition Approach Experiments & Analysis
How Local and Global Weights Can Help
When computing the attention matrix, POLAR-LOC only useslocal weights while POLAR-GLO only global weights.
Aminer76
77
78
79
80
81
NDCG
@3
Patent
60
62
64
66
68
RARD43.0
43.5
44.0
44.5
45.0
45.5
46.0
46.5
47.0POLAR-LOCPOLAR-GLOPOLAR-ALL
Figure: The performance of different attention matrices
16 / 20
![Page 29: POLAR: Attention-based CNN for One-shot Personalized ... · 1Vinyals et al., Matching Networks for One Shot Learning. 6/20. Introduction & Problem De nitionApproachExperiments & Analysis](https://reader034.vdocuments.site/reader034/viewer/2022050409/5f85e49ba1d3a8189b46dba2/html5/thumbnails/29.jpg)
Introduction & Problem Definition Approach Experiments & Analysis
Case Study: How Local and Global Weights work?
0 2 4 6 8
0
2
4
6
8
matching matrix
0 2 4 6 8
0
2
4
6
8
global weights
0 2 4 6 8
0
2
4
6
8
local weights
0 2 4 6 8
0
2
4
6
8
weighted matching matrix
Figure: The visualization result of four matrices used in the matching of apair of texts. The brighter the pixel is, the larger value it has.T1:novel robust stability criteria (for) stochastic hopfield neural networks(with) time delays.T2:new delay dependent stability criteria (for) neural networks (with)time varying delay
17 / 20
![Page 30: POLAR: Attention-based CNN for One-shot Personalized ... · 1Vinyals et al., Matching Networks for One Shot Learning. 6/20. Introduction & Problem De nitionApproachExperiments & Analysis](https://reader034.vdocuments.site/reader034/viewer/2022050409/5f85e49ba1d3a8189b46dba2/html5/thumbnails/30.jpg)
Introduction & Problem Definition Approach Experiments & Analysis
Case Study: How Local and Global Weights work?
Table: The statistical analysis of the local and global weights
Weight Max Min Mean Std
Local 2.00 1.00 1.20 0.15
Global 1.96 1.08 1.86 0.08
17 / 20
![Page 31: POLAR: Attention-based CNN for One-shot Personalized ... · 1Vinyals et al., Matching Networks for One Shot Learning. 6/20. Introduction & Problem De nitionApproachExperiments & Analysis](https://reader034.vdocuments.site/reader034/viewer/2022050409/5f85e49ba1d3a8189b46dba2/html5/thumbnails/31.jpg)
Introduction & Problem Definition Approach Experiments & Analysis
Sensitivity Analysis of Hyperparameters
0 0.25 0.5 1 2 4local weight coefficient
86.0
86.5
87.0
87.5
88.0
88.5
89.0
89.5
NDCG
@10
0 1/8 1/4 1/2 1global weight coefficient
86.0
86.5
87.0
87.5
88.0
88.5
89.0
89.5
NDCG
@10
Figure: Performance comparison for POLAR-LOC with different αs andPOLAR-GLO with different βs on the AMiner dataset
18 / 20
![Page 32: POLAR: Attention-based CNN for One-shot Personalized ... · 1Vinyals et al., Matching Networks for One Shot Learning. 6/20. Introduction & Problem De nitionApproachExperiments & Analysis](https://reader034.vdocuments.site/reader034/viewer/2022050409/5f85e49ba1d3a8189b46dba2/html5/thumbnails/32.jpg)
Introduction & Problem Definition Approach Experiments & Analysis
Conclusion
We define the problem of one-shot personalized articlerecommendation.
We utilize the framework of one-shot learning to deal with thesparse user feedback and propose an attention-based CNN fortext similarity.
We conduct experiments, whose results prove the effectivenessof the proposed model.
19 / 20
![Page 33: POLAR: Attention-based CNN for One-shot Personalized ... · 1Vinyals et al., Matching Networks for One Shot Learning. 6/20. Introduction & Problem De nitionApproachExperiments & Analysis](https://reader034.vdocuments.site/reader034/viewer/2022050409/5f85e49ba1d3a8189b46dba2/html5/thumbnails/33.jpg)
Introduction & Problem Definition Approach Experiments & Analysis
Any Questions?
20 / 20