Student research highlight: Secure and resilient distributed machine learning under adversarial environments

Rui Zhang, Quanyan Zhu

Research output: Contribution to journalReview article

Abstract

Machine learning algorithms, such as support vector machines (SVMs), neutral networks, and decision trees (DTs) have been widely used in data processing for estimation and detection. They can be used to classify samples based on a model built from training data. However, under the assumption that training and testing samples come from the same natural distribution, an attacker who can generate or modify training data will lead to misclassification or misestimation. For example, a spam filter will fail to recognize input spam messages after training crafted data provided by attackers [1].

Original languageEnglish (US)
Article number7478408
Pages (from-to)34-36
Number of pages3
JournalIEEE Aerospace and Electronic Systems Magazine
Volume31
Issue number3
DOIs
StatePublished - Mar 1 2016

Fingerprint

machine learning
Decision trees
students
Learning algorithms
Support vector machines
Learning systems
education
student
Students
Testing
messages
filter
filters

ASJC Scopus subject areas

  • Electrical and Electronic Engineering
  • Aerospace Engineering
  • Space and Planetary Science

Cite this

Student research highlight : Secure and resilient distributed machine learning under adversarial environments. / Zhang, Rui; Zhu, Quanyan.

In: IEEE Aerospace and Electronic Systems Magazine, Vol. 31, No. 3, 7478408, 01.03.2016, p. 34-36.

Research output: Contribution to journalReview article

@article{7e8f2ca241324199b67bb4f3e22e00fb,
title = "Student research highlight: Secure and resilient distributed machine learning under adversarial environments",
abstract = "Machine learning algorithms, such as support vector machines (SVMs), neutral networks, and decision trees (DTs) have been widely used in data processing for estimation and detection. They can be used to classify samples based on a model built from training data. However, under the assumption that training and testing samples come from the same natural distribution, an attacker who can generate or modify training data will lead to misclassification or misestimation. For example, a spam filter will fail to recognize input spam messages after training crafted data provided by attackers [1].",
author = "Rui Zhang and Quanyan Zhu",
year = "2016",
month = "3",
day = "1",
doi = "10.1109/MAES.2016.150202",
language = "English (US)",
volume = "31",
pages = "34--36",
journal = "IEEE Aerospace and Electronic Systems Magazine",
issn = "0885-8985",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
number = "3",

}

TY - JOUR

T1 - Student research highlight

T2 - Secure and resilient distributed machine learning under adversarial environments

AU - Zhang, Rui

AU - Zhu, Quanyan

PY - 2016/3/1

Y1 - 2016/3/1

N2 - Machine learning algorithms, such as support vector machines (SVMs), neutral networks, and decision trees (DTs) have been widely used in data processing for estimation and detection. They can be used to classify samples based on a model built from training data. However, under the assumption that training and testing samples come from the same natural distribution, an attacker who can generate or modify training data will lead to misclassification or misestimation. For example, a spam filter will fail to recognize input spam messages after training crafted data provided by attackers [1].

AB - Machine learning algorithms, such as support vector machines (SVMs), neutral networks, and decision trees (DTs) have been widely used in data processing for estimation and detection. They can be used to classify samples based on a model built from training data. However, under the assumption that training and testing samples come from the same natural distribution, an attacker who can generate or modify training data will lead to misclassification or misestimation. For example, a spam filter will fail to recognize input spam messages after training crafted data provided by attackers [1].

UR - http://www.scopus.com/inward/record.url?scp=84976566521&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84976566521&partnerID=8YFLogxK

U2 - 10.1109/MAES.2016.150202

DO - 10.1109/MAES.2016.150202

M3 - Review article

AN - SCOPUS:84976566521

VL - 31

SP - 34

EP - 36

JO - IEEE Aerospace and Electronic Systems Magazine

JF - IEEE Aerospace and Electronic Systems Magazine

SN - 0885-8985

IS - 3

M1 - 7478408

ER -