Filters








29 Hits in 5.3 sec

Transcending Transcend: Revisiting Malware Classification in the Presence of Concept Drift [article]

Federico Barbero, Feargus Pendlebury, Fabio Pierazzi, Lorenzo Cavallaro
2021 arXiv   pre-print
One promising method to cope with concept drift is classification with rejection in which examples that are likely to be misclassified are instead quarantined until they can be expertly analyzed.  ...  Machine learning for malware classification shows encouraging results, but real deployments suffer from performance degradation as malware authors adapt their techniques to evade detection.  ...  The impetus for concept drift in malware classification is the adversarial nature of the task.  ... 
arXiv:2010.03856v5 fatcat:227lvueaobf6npmjz5zu2vmjwy

Machine Learning (In) Security: A Stream of Problems [article]

Fabrício Ceschin and Heitor Murilo Gomes and Marcus Botacin and Albert Bifet and Bernhard Pfahringer and Luiz S. Oliveira and André Grégio
2020 arXiv   pre-print
In this work, we list, detail, and discuss some of the challenges of applying ML to cybersecurity, including concept drift, concept evolution, delayed labels, and adversarial machine learning.  ...  One of these challenges is the concept drift, that actually creates an arms race between attackers and defenders, given that any attacker may create novel, different threats as time goes by (to overcome  ...  Jordaney et al. presented Transcend, a framework to identify concept drift in classification models which compares the samples used to train the models with those seen during deployment [83] .  ... 
arXiv:2010.16045v1 fatcat:edph3d2f7zat3jl4bjyvawbvty

DAEMON: Dataset/Platform-Agnostic Explainable Malware Classification Using Multi-Stage Feature Mining

Ron Korine, Danny Hendler
2021 IEEE Access  
[67] proposes novel measures for detecting concept drift that can significantly reduce the computational cost of Transcend.  ...  Transcend effectively identifies concept drifts in both binary and multi-class classifiers. Barbero et al.  ...  DAEMON obtained an excellent classification accuracy of 99.72% in a 5-fold cross validation applied to Microsoft's training set and came out 3rd in terms of logloss out of more than 370 different classifiers  ... 
doi:10.1109/access.2021.3082173 fatcat:bccfewzkprghhmxnhkncjblcde

Learning Fast and Slow: PROPEDEUTICA for Real-time Malware Detection [article]

Ruimin Sun, Xiaoyong Yuan, Pan He, Qile Zhu, Aokun Chen, Andre Gregio, Daniela Oliveira, Xiaolin Li
2021 arXiv   pre-print
Existing malware detectors on safety-critical devices have difficulties in runtime detection due to the performance overhead.  ...  In this paper, we introduce PROPEDEUTICA, a framework for efficient and effective real-time malware detection, leveraging the best of conventional machine learning (ML) and deep learning (DL) techniques  ...  To cope with the fast-evolving nature of malware, Transcend [103] addresses concept drift in malware classification. Transcend can detect aging machine learning models before their degradation.  ... 
arXiv:1712.01145v2 fatcat:mnmb4mpidbc5hn44ajdec7ssga

A Review of Android Malware Detection Approaches based on Machine Learning

Kaijun Liu, Shengwei Xu, Guoai Xu, Miao Zhang, Dawei Sun, Haifeng Liu
2020 IEEE Access  
Android applications are developing rapidly across the mobile ecosystem, but Android malware is also emerging in an endless stream.  ...  We briefly introduce some background on Android applications, including the Android system architecture, security mechanisms, and classification of Android malware.  ...  The Transcend framework [293] utilizes statistical methods to detect concept drift and is not subject to the machine learning algorithm used by the classifier; however, it does not propose a specific  ... 
doi:10.1109/access.2020.3006143 fatcat:5rn2qg67ezdixkrefwxmyejhsi

Lessons Learnt on Reproducibility in Machine Learning Based Android Malware Detection

Nadia Daoudi, Kevin Allix, Tegawendé F. Bissyandé, Jacques Klein
2021 Empirical Software Engineering  
thus becomes paramount despite the little interest such mundane and practical aspects seem to attract in the malware detection field.  ...  In this respect, recently, research achievements on machine learning based malware detection are being considered for thorough evaluation by the community.  ...  The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material.  ... 
doi:10.1007/s10664-021-09955-7 fatcat:syqllmgmhfbx7itj5sf3eixlha

Machine Learning Methods for Computer Security (Dagstuhl Perspectives Workshop 12371)

Anthony D. Joseph, Pavel Laskov, Fabio Roli, J. Doug Tygar, Blaine Nelson, Marc Herbstritt
2013 Dagstuhl Reports  
The open problems identified in the field ranged from traditional applications of machine learning in security, such as attack detection and analysis of malicious software, to methodological issues related  ...  Examples of such applications are social media spam, plagiarism detection, authorship identification, copyright enforcement, computer vision (particularly in the context of biometrics), and sentiment analysis  ...  An alternative approach based on supervised learning enabled classification of malware into known families as well as detection of novel malware strains [99] .  ... 
doi:10.4230/dagrep.2.9.109 dblp:journals/dagstuhl-reports/JosephLRTN12 fatcat:4x3ng2szxfg5jnkf5rtwsmttrm

Cloud Computing and Security in the IoT Era

Yahya Abssi, Shailendra Mishra, Manoj Kumar Shukla
2020 Helix  
The study proposes numerous solutions for the challenges of privacy and security in the cloud and IoT. The paper provides an in-depth examination of security challenges in cloud computing and IoT.  ...  The stored information in clouds can be manipulated without the need for extensive physical activities.  ...  The work in this paper is organized in a different section, an introduction gives the overall concept of research to be conducted, background and related work are discussed in second.  ... 
doi:10.29042/2020-10-4-51-58 fatcat:2og6y7w2yzhlvcpy7a7ycgncd4

Truth Will Out

Wissam Aoudi, Mikel Iturbe, Magnus Almgren
2018 Proceedings of the 2018 ACM SIGSAC Conference on Computer and Communications Security - CCS '18  
In this paper, we present pasad, a novel stealthy-attack detection mechanism that monitors time series of sensor measurements in real time for structural changes in the process behavior.  ...  Experimental results show that pasad is capable of detecting not only significant deviations in the process behavior, but also subtle attack-indicating changes, significantly raising the bar for strategic  ...  We describe the procedure in detail in §2.3, but first we introduce preliminary concepts in linear algebra.  ... 
doi:10.1145/3243734.3243781 dblp:conf/ccs/AoudiIA18 fatcat:rh3bvs2wcbgdli3nngjohkv42e

Between Hype and Understatement: Reassessing Cyber Risks as a Security Strategy

Audrey Guinchard
2011 Journal of Strategic Security  
Because these vulnerabilities that exist in the network are not themselves illegal, they tend to be overlooked in the debate on cyber security.  ...  Most of the actions that fall under the trilogy of cyber crime, terrorism, and war exploit pre-existing weaknesses in the underlying technology.  ...  In that respect, the IC3's 79 categories for classification of cyber incidents robustly covers the different types of cyber crimes.  ... 
doi:10.5038/1944-0472.4.2.5 fatcat:cqfwfgitprgnrfyjin2rn7b3ri

European Baseline Report On Current Oc/Tn Specifics And Collection Of Sources

Pablo Martín Rodríguez
2017 Zenodo  
into account for modelling purposes.  ...  This review will allow extracting in Section 4 some relevant conclusions as to the methodological framework where the subsequent development of the model that the TAKEDOWN Project will create and apply  ...  the so-called seduction or drift model (Goldsmith/Brewer, 2015) .  ... 
doi:10.5281/zenodo.1232126 fatcat:7qu2p72ocbbkbghr7q2apzqco4

Machine Learning Methods for Computer Security (Dagstuhl Perspectives Workshop 12371)

Anthony Joseph, Pavel Laskov, Fabio Roli, J Tygar, Blaine Nelson
unpublished
We thank all participants and, in particular, Mark Braunstein and Peter Walden for many valuable comments and suggestions regarding this text.  ...  Marc Herbstritt, and Jutka Gasiorowski for their help in organizing this workshop and preparing this report. Participants Jonas Acknowledgement.  ...  An alternative approach based on supervised learning enabled classification of malware into known families as well as detection of novel malware strains [99] .  ... 
fatcat:2mmdlhygcvb2tfzbbez7i4de4u

Net neutrality - How relevant is it to Australia?

James Endres
2009 Telecommunications Journal of Australia  
Moreover, infringements of Australian classification are unlikely to be detectable by the Classification Board.  ...  operators in Japan regularly introduce new handset models in Spring and Autumn/Winter with occasional minor model changes in Summer. 12 The author express thanks to Mr.  ...  In particular in Telstra's case that includes big issues like the conception of what is the purpose of a large corporation (nowadays reduced to 'creating shareholder value'), the change in emphasis of  ... 
doi:10.2104/tja09022 fatcat:iyhfeahrmverdoaskclswtokx4

IAFS 2017 Abstracts

2017 Forensic Science International  
THE ESTIMATION OF THE POSTMORTEM INTERVAL: A PHYSICOCHEMICAL MODEL  ...  ASSOCIATION BETWEEN SCN5A GENE AND SUDDEN UNEXPLAINED NOCTURNAL DEATH SYNDROME IN THAI CADAVERS Disclosure: All authors have declared no conflicts of interest.  ...  The overall decrease in RMSE was detected in combined model as compared toTMD (0.03-0.06) andTME (0.2-0.8).  ... 
doi:10.1016/j.forsciint.2017.07.019 pmid:28743352 fatcat:pdrympkytzbzlkah4vnqismhau

IEEE Reliability Society Technical Operations Annual Technology Report for 2008

2009 IEEE Transactions on Reliability  
The ethics cluster concepts showed up selectively in all domains, but tended to show an exceedingly low correlation with concepts dealing exclusively with automation and derivatives.  ...  models to detect malicious code that takes into account both static and dynamic analysis techniques, and to formulate the model such that it is cost effective and highly accurate. 2) To develop better  ... 
doi:10.1109/tr.2009.2020845 fatcat:mt4ppqhpyrcbbkofg5bbjinrie
« Previous Showing results 1 — 15 out of 29 results