Detection under privileged information

Z. Berkay Celik, Patrick McDaniel, Rauf Izmailov, Nicolas Papernot, Ryan Sheatsley, Raquel Alvarez, Ananthram Swami

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

For well over a quarter century, detection systems have been driven by models learned from input features collected from real or simulated environments. An artifact (e.g., network event, potential malware sample, suspicious email) is deemed malicious or non-malicious based on its similarity to the learned model at runtime. However, the training of the models has been historically limited to only those features available at runtime. In this paper, we consider an alternate learning approach that trains models using “privileged” information-features available at training time but not at runtime-to improve the accuracy and resilience of detection systems. In particular, we adapt and extend recent advances in knowledge transfer, model influence, and distillation to enable the use of forensic or other data unavailable at runtime in a range of security domains. An empirical evaluation shows that privileged information increases precision and recall over a system with no privileged information: we observe up to 7.7% relative decrease in detection error for fast-flux bot detection, 8.6% for malware traffic detection, 7.3% for malware classification, and 16.9% for face recognition. We explore the limitations and applications of different privileged information techniques in detection systems. Such techniques provide a new means for detection systems to learn from data that would otherwise not be available at runtime.

Original languageEnglish (US)
Title of host publicationASIACCS 2018 - Proceedings of the 2018 ACM Asia Conference on Computer and Communications Security
PublisherAssociation for Computing Machinery, Inc
Pages199-206
Number of pages8
ISBN (Electronic)9781450355766
DOIs
StatePublished - May 29 2018
Event13th ACM Symposium on Information, Computer and Communications Security, ASIACCS 2018 - Incheon, Korea, Republic of
Duration: Jun 4 2018Jun 8 2018

Publication series

NameASIACCS 2018 - Proceedings of the 2018 ACM Asia Conference on Computer and Communications Security

Other

Other13th ACM Symposium on Information, Computer and Communications Security, ASIACCS 2018
CountryKorea, Republic of
CityIncheon
Period6/4/186/8/18

Fingerprint

Error detection
Electronic mail
Face recognition
Distillation
Fluxes
Malware

All Science Journal Classification (ASJC) codes

  • Software
  • Computer Science Applications
  • Information Systems
  • Computer Networks and Communications

Cite this

Berkay Celik, Z., McDaniel, P., Izmailov, R., Papernot, N., Sheatsley, R., Alvarez, R., & Swami, A. (2018). Detection under privileged information. In ASIACCS 2018 - Proceedings of the 2018 ACM Asia Conference on Computer and Communications Security (pp. 199-206). (ASIACCS 2018 - Proceedings of the 2018 ACM Asia Conference on Computer and Communications Security). Association for Computing Machinery, Inc. https://doi.org/10.1145/3196494.3196502
Berkay Celik, Z. ; McDaniel, Patrick ; Izmailov, Rauf ; Papernot, Nicolas ; Sheatsley, Ryan ; Alvarez, Raquel ; Swami, Ananthram. / Detection under privileged information. ASIACCS 2018 - Proceedings of the 2018 ACM Asia Conference on Computer and Communications Security. Association for Computing Machinery, Inc, 2018. pp. 199-206 (ASIACCS 2018 - Proceedings of the 2018 ACM Asia Conference on Computer and Communications Security).
@inproceedings{ac97a5ff27e64a869617d5ede23cce18,
title = "Detection under privileged information",
abstract = "For well over a quarter century, detection systems have been driven by models learned from input features collected from real or simulated environments. An artifact (e.g., network event, potential malware sample, suspicious email) is deemed malicious or non-malicious based on its similarity to the learned model at runtime. However, the training of the models has been historically limited to only those features available at runtime. In this paper, we consider an alternate learning approach that trains models using “privileged” information-features available at training time but not at runtime-to improve the accuracy and resilience of detection systems. In particular, we adapt and extend recent advances in knowledge transfer, model influence, and distillation to enable the use of forensic or other data unavailable at runtime in a range of security domains. An empirical evaluation shows that privileged information increases precision and recall over a system with no privileged information: we observe up to 7.7{\%} relative decrease in detection error for fast-flux bot detection, 8.6{\%} for malware traffic detection, 7.3{\%} for malware classification, and 16.9{\%} for face recognition. We explore the limitations and applications of different privileged information techniques in detection systems. Such techniques provide a new means for detection systems to learn from data that would otherwise not be available at runtime.",
author = "{Berkay Celik}, Z. and Patrick McDaniel and Rauf Izmailov and Nicolas Papernot and Ryan Sheatsley and Raquel Alvarez and Ananthram Swami",
year = "2018",
month = "5",
day = "29",
doi = "10.1145/3196494.3196502",
language = "English (US)",
series = "ASIACCS 2018 - Proceedings of the 2018 ACM Asia Conference on Computer and Communications Security",
publisher = "Association for Computing Machinery, Inc",
pages = "199--206",
booktitle = "ASIACCS 2018 - Proceedings of the 2018 ACM Asia Conference on Computer and Communications Security",

}

Berkay Celik, Z, McDaniel, P, Izmailov, R, Papernot, N, Sheatsley, R, Alvarez, R & Swami, A 2018, Detection under privileged information. in ASIACCS 2018 - Proceedings of the 2018 ACM Asia Conference on Computer and Communications Security. ASIACCS 2018 - Proceedings of the 2018 ACM Asia Conference on Computer and Communications Security, Association for Computing Machinery, Inc, pp. 199-206, 13th ACM Symposium on Information, Computer and Communications Security, ASIACCS 2018, Incheon, Korea, Republic of, 6/4/18. https://doi.org/10.1145/3196494.3196502

Detection under privileged information. / Berkay Celik, Z.; McDaniel, Patrick; Izmailov, Rauf; Papernot, Nicolas; Sheatsley, Ryan; Alvarez, Raquel; Swami, Ananthram.

ASIACCS 2018 - Proceedings of the 2018 ACM Asia Conference on Computer and Communications Security. Association for Computing Machinery, Inc, 2018. p. 199-206 (ASIACCS 2018 - Proceedings of the 2018 ACM Asia Conference on Computer and Communications Security).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - Detection under privileged information

AU - Berkay Celik, Z.

AU - McDaniel, Patrick

AU - Izmailov, Rauf

AU - Papernot, Nicolas

AU - Sheatsley, Ryan

AU - Alvarez, Raquel

AU - Swami, Ananthram

PY - 2018/5/29

Y1 - 2018/5/29

N2 - For well over a quarter century, detection systems have been driven by models learned from input features collected from real or simulated environments. An artifact (e.g., network event, potential malware sample, suspicious email) is deemed malicious or non-malicious based on its similarity to the learned model at runtime. However, the training of the models has been historically limited to only those features available at runtime. In this paper, we consider an alternate learning approach that trains models using “privileged” information-features available at training time but not at runtime-to improve the accuracy and resilience of detection systems. In particular, we adapt and extend recent advances in knowledge transfer, model influence, and distillation to enable the use of forensic or other data unavailable at runtime in a range of security domains. An empirical evaluation shows that privileged information increases precision and recall over a system with no privileged information: we observe up to 7.7% relative decrease in detection error for fast-flux bot detection, 8.6% for malware traffic detection, 7.3% for malware classification, and 16.9% for face recognition. We explore the limitations and applications of different privileged information techniques in detection systems. Such techniques provide a new means for detection systems to learn from data that would otherwise not be available at runtime.

AB - For well over a quarter century, detection systems have been driven by models learned from input features collected from real or simulated environments. An artifact (e.g., network event, potential malware sample, suspicious email) is deemed malicious or non-malicious based on its similarity to the learned model at runtime. However, the training of the models has been historically limited to only those features available at runtime. In this paper, we consider an alternate learning approach that trains models using “privileged” information-features available at training time but not at runtime-to improve the accuracy and resilience of detection systems. In particular, we adapt and extend recent advances in knowledge transfer, model influence, and distillation to enable the use of forensic or other data unavailable at runtime in a range of security domains. An empirical evaluation shows that privileged information increases precision and recall over a system with no privileged information: we observe up to 7.7% relative decrease in detection error for fast-flux bot detection, 8.6% for malware traffic detection, 7.3% for malware classification, and 16.9% for face recognition. We explore the limitations and applications of different privileged information techniques in detection systems. Such techniques provide a new means for detection systems to learn from data that would otherwise not be available at runtime.

UR - http://www.scopus.com/inward/record.url?scp=85049163150&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85049163150&partnerID=8YFLogxK

U2 - 10.1145/3196494.3196502

DO - 10.1145/3196494.3196502

M3 - Conference contribution

AN - SCOPUS:85049163150

T3 - ASIACCS 2018 - Proceedings of the 2018 ACM Asia Conference on Computer and Communications Security

SP - 199

EP - 206

BT - ASIACCS 2018 - Proceedings of the 2018 ACM Asia Conference on Computer and Communications Security

PB - Association for Computing Machinery, Inc

ER -

Berkay Celik Z, McDaniel P, Izmailov R, Papernot N, Sheatsley R, Alvarez R et al. Detection under privileged information. In ASIACCS 2018 - Proceedings of the 2018 ACM Asia Conference on Computer and Communications Security. Association for Computing Machinery, Inc. 2018. p. 199-206. (ASIACCS 2018 - Proceedings of the 2018 ACM Asia Conference on Computer and Communications Security). https://doi.org/10.1145/3196494.3196502