Document Detail

Conditional Anomaly Detection with Soft Harmonic Functions.
MedLine Citation:
PMID:  25309142     Owner:  NLM     Status:  Publisher    
In this paper, we consider the problem of conditional anomaly detection that aims to identify data instances with an unusual response or a class label. We develop a new non-parametric approach for conditional anomaly detection based on the soft harmonic solution, with which we estimate the confidence of the label to detect anomalous mislabeling. We further regularize the solution to avoid the detection of isolated examples and examples on the boundary of the distribution support. We demonstrate the efficacy of the proposed method on several synthetic and UCI ML datasets in detecting unusual labels when compared to several baseline approaches. We also evaluate the performance of our method on a real-world electronic health record dataset where we seek to identify unusual patient-management decisions.
Michal Valko; Branislav Kveton; Hamed Valizadegan; Gregory F Cooper; Milos Hauskrecht
Publication Detail:
Journal Detail:
Title:  Proceedings / IEEE International Conference on Data Mining. IEEE International Conference on Data Mining     Volume:  2011     ISSN:  1550-4786     ISO Abbreviation:  Proc IEEE Int Conf Data Min     Publication Date:  2011  
Date Detail:
Created Date:  2014-10-13     Completed Date:  -     Revised Date:  2014-10-16    
Medline Journal Info:
Nlm Unique ID:  101529961     Medline TA:  Proc IEEE Int Conf Data Min     Country:  -    
Other Details:
Languages:  ENG     Pagination:  735-743     Citation Subset:  -    
Export Citation:
APA/MLA Format     Download EndNote     Download BibTex
MeSH Terms
Grant Support
R01 GM088224/GM/NIGMS NIH HHS; R01 GM088224-01/GM/NIGMS NIH HHS; R01 GM088224-02/GM/NIGMS NIH HHS; R01 GM088224-03/GM/NIGMS NIH HHS; R01 LM010019/LM/NLM NIH HHS; R01 LM010019-01A1/LM/NLM NIH HHS; R01 LM010019-02/LM/NLM NIH HHS; R01 LM010019-03/LM/NLM NIH HHS

From MEDLINE®/PubMed®, a database of the U.S. National Library of Medicine

Previous Document:  Learning classification with auxiliary probabilistic information.
Next Document:  Neonatal abstinence syndrome: essentials for the practitioner.