2018 IEEE International Conference on Industrial Engineering and Engineering Management (IEEM) 2018
DOI: 10.1109/ieem.2018.8607671
|View full text |Cite
|
Sign up to set email alerts
|

Study on Unbalanced Binary Classification with Unknown Misclassification Costs

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
2
1
1
1

Citation Types

0
5
0

Year Published

2020
2020
2024
2024

Publication Types

Select...
4
1
1
1

Relationship

0
7

Authors

Journals

citations
Cited by 8 publications
(5 citation statements)
references
References 6 publications
0
5
0
Order By: Relevance
“…We use the ADAM optimizer 47 and a loss function of binary cross entropy with logits loss 48 . We find that giving a higher weight to learning positive predictions stabilizes training and accelerates convergence 49 .…”
Section: Training Model By the Rough Manually Labeled Datamentioning
confidence: 80%
“…We use the ADAM optimizer 47 and a loss function of binary cross entropy with logits loss 48 . We find that giving a higher weight to learning positive predictions stabilizes training and accelerates convergence 49 .…”
Section: Training Model By the Rough Manually Labeled Datamentioning
confidence: 80%
“…We use the ADAM optimizer 48 and a loss function of binary cross entropy with logits loss 49 . We have determined that giving a higher weight to learning positive predictions stabilizes training and accelerates convergence 50 , stabilizes training and accelerates convergence 50 . , where white circles in (a) correspond to the manual labeling and red dots in (b) to the prediction of the trained model.…”
Section: Training Model By the Rough Manually Labeled Datamentioning
confidence: 99%
“…The equal class distribution in the target variable is a critical issue to avoid the miss-classification problem since fraud transactions are uncommon in most datasets. ML algorithms can easily find their regularities in large datasets compared to small datasets [25]. The overlapped attribute values also create problems in a large number of transactions.…”
Section: Literature Reviewmentioning
confidence: 99%