Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • harvard1
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
Unsupervised Deep Learning to Mitigate Operational Risk through Anomaly Detection in Big Data
KTH, School of Electrical Engineering and Computer Science (EECS).
KTH, School of Electrical Engineering and Computer Science (EECS).
2018 (English)Independent thesis Basic level (degree of Bachelor), 10 credits / 15 HE creditsStudent thesis
Abstract [en]

Exposure to operational risk due to bad data quality is a ubiquitous problem for the incumbent financial industry. This study explores the possibility of using unsupervised machine learning, and more specifically, deep learning to identify anomalies in large, unlabeled datasets as an attempt to mitigate operational risk by improving data quality. The algorithm used is a sparse autoencoder neural network. The three different datasets of of approximately one million data points each that were used for this research was provided by a large Swedish bank, Handelsbanken AB. Handelsbanken was also responsible for the insertion of between 500 to 2000 intentional, potential damage inflicting errors in each respective dataset. As part of this study a neural network has been constructed for the purpose for of finding the inserted errors. In two out of the three datasets, 100% of the errors could be found, in the third around 70% of errors could be found. This implies the elimination of operational risk due to bad data quality in two out of three datasets and significantly mitigating its occurrence in the last dataset. The study proves the concept that deep learning can be used to mitigate operational risk and elaborates on the potential impact and practical value of this solution. Examples of such impact could be more precise decision making, better customer service and less exposure to operational and reputation risk. The algorithm is potentially useful for any organization that relies on data to conduct their operations.

Place, publisher, year, edition, pages
2018. , p. 12
Series
TRITA-EECS-EX ; 2018:438
Keywords [en]
Deep Learning, Neural Network, Autoencoder, Operational Risk, Data Quality, Unsupervised Learning
National Category
Computer and Information Sciences
Identifiers
URN: urn:nbn:se:kth:diva-250232OAI: oai:DiVA.org:kth-250232DiVA, id: diva2:1307240
External cooperation
Handelsbanken
Educational program
Master of Science in Engineering - Industrial Engineering and Management
Supervisors
Examiners
Available from: 2019-05-13 Created: 2019-04-26 Last updated: 2019-05-13Bibliographically approved

Open Access in DiVA

No full text in DiVA

By organisation
School of Electrical Engineering and Computer Science (EECS)
Computer and Information Sciences

Search outside of DiVA

GoogleGoogle Scholar

urn-nbn

Altmetric score

urn-nbn
Total: 180 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • harvard1
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf