Postdoc in Adversarial Machine Learning for Cyber-Security

onsdag 03 jul 19

Send ansøgning

Frist 26. august 2019
Du kan søge om jobbet ved DTU Compute ved at udfylde den efterfølgende ansøgningsformular.

Ansøg online

The Cyber Security Section at DTU Compute conducts interdisciplinary research in most aspects of computer and information security, ranging from advanced cryptography and access control mechanisms, for ensuring secrecy and authentication in todays’ communications, to the design, analysis and implementation of novel protection mechanisms for enhanced security, privacy, trustworthiness, and reliability in numerous emerging IoT applications. Through methods such as applied cryptography, hardware enhanced security, modelling and security analysis of systems, and verification, we are building security into new technology, providing assurance to the user that they are interacting with a trusted platform. Our research focuses on cryptography; trusted computing; secure communications; privacy and authentication; and security verification – with applications in, e.g., automotive, future Internet and 5G, smart grid, healthcare, e Payment, etc.

Research field of the position
This position is part of the research project SecDNS (“Identification of Cyber-Security Threats using Machine Learning”) funded by the Innovation Fund Denmark programme. The goal of SecDNS is to engage artificial intelligence and data science technologies towards developing a unified adversarial classification framework for identifying complex cyber-security threats in the Internet (e.g., malicious domains) and other cloud-based networking paradigms; taking into account uncertainty of data provenance, used for the classification, while handling the necessary belief inference and propagation modelling. The use of machine learning models has become ubiquitous. Their predictions are used to make decisions about a number of critical applications including (amongst others) security in identifying complex cyber-security threats in the context of the Internet (e.g., malicious domains) and other networking environments. These models are widely used in many cyber defence systems for network security operations, malware analysis, etc. But despite the many successes, the very property that makes machine learning desirable: adaptability, is a vulnerability that may be exploited by an attacker that could potentially result in the severe degradation of the integrity, security and performance of cyber defence systems.

The successful candidate will join the Cyber Security Section working on the delivery of new algorithmic and theoretical results on the robustness of machine learning methods, including also reinforcement learning, in adversarial environments, and the development of appropriate classification models capable of handling the uncertainty of the data correctness used for training. Furthermore, carrying out research towards the specification of appropriate belief inference and propagation mechanisms based on a detailed feature constraints analysis and trust management of all data sources as well as market landscapes to allow a detailed design phase to proceed. 

Responsibilities and tasks
The goal of the postdoc is to undertake a range of research activities within adversarial machine learning and to investigate, understand, evaluate and improve the effectiveness of machine learning, inference and propagation methods in the presence of motivated and sophisticated adversaries. Carrying out research into the technical building blocks of the proposed architecture including machine learning, reinforcement learning, trust management, Bayesian inference and subjective logic for providing a unified adversarial classification framework.

We are looking for an excellent, motivated, self-driven post-doctoral candidate to conduct high-quality research within the following key technological aspects:

  • the development of optimal defense strategies for improving the security of existing machine learning algorithms in the context of cyber-security. This will involve studying potential attacks, such as data poisoning, developing targeted defenses against these attacks, and efficient algorithms for computing the optimal defense strategy;
  • the specification and development of appropriate belief inference and propagation models (e.g., Bayesian Inference, Trust Management, Dempster-Shafer Theory, etc.) capable of handling various levels of uncertainty of the different data features used in the classification models. Belief propagation will be based on a hybrid approach integration the most appropriate of the investigated inference techniques;
  • the security validation of the implemented mechanisms and protocols. This comprised of all activities that aim at demonstrating the security qualities of the specified protocols. Hence, it includes verification, analysis, testing, and performance evaluation;
In addition to supporting the research project with internal and external collaborators, the position will serve as a platform for the research fellow to develop their career and profile as an independent researcher. The potential for the development of knowledge and skills at the intersection of trusted computing and cloud computing may help the research fellow develop an independent research agenda.

The applicant will work in a team of PhD students, postdocs and faculty members in the section, and must contribute with research towards the overall goals of the SecDNS project. The applicant is expected to interact with our collaborators on the design of secure systems based on the use of trusted computing technologies. The applicant is also expected to give limited contributions to teaching and training activities as well as supervision of students.

Candidates should hold, or are near completion of, a PhD in a relevant subject. (Relevant professional experience may be considered). The successful applicant will need to demonstrate: (i) experience and knowledge of machine learning methods including reinforcement and deep learning and optimization techniques, (ii) strong background in the theory and practice of security and fault tolerance in intrusion detection systems, and (ii) solid foundation in linear algebra for the belied inference modelling. Good implementation skills and practical experience are also desirable. Furthermore, good command of the English language is essential.

Successful candidates have a unique opportunity of participating in the shaping of challenging and innovative research themes, as well as contributing to the ambitious research agenda of the DTU Cyber Security section supported by a number of national and international grants. They will be given excellent conditions for the development of their research skills, in terms of working conditions, mentoring and lab facilities.

We offer
DTU is a leading technical university globally recognized for the excellence of its research, education, innovation and scientific advice. We offer a rewarding and challenging job in an international environment. We strive for academic excellence in an environment characterized by collegial respect and academic freedom tempered by responsibility.

Salary and terms of employment
The appointment will be based on the collective agreement with the Danish Confederation of Professional Associations. The allowance will be agreed upon with the relevant union.

The duration of the position is 2 years, and we aim for at starting date of 1 October 2019 or as soon as possible after that.

You can read about career paths at DTU 

Further information
Further information concerning the project can be obtained from Christian D. Jensen, mail, tel. +45 4525 3724) or Athanasios Giannetsos, mail, tel +45 4525 3009, in the Cyber Security Section at DTU Compute.

You can read more about DTU compute at   

Application procedure
Please submit your application no later than  26 August 2019 (local time). Applications must be submitted as one PDF file containing all materials to be given consideration. To apply, please open the link "Apply online", fill out the online application form, and attach all your materials in English in one PDF file. The file must include: 
  • Application (cover letter)
  • CV
  • Diploma (MSc/PhD)
  • List of publications
  • Research Statement including a brief description of how past experience and future plans fit with the advertised position (max 2 pages)
Applications and enclosures received after the deadline will not be considered.

All interested candidates irrespective of age, gender, disability, race, religion or ethnic background are encouraged to apply.

DTU Compute has a total staff of 400 including 100 faculty members and 130 PhD students. We offer introductory courses in mathematics, statistics, and computer science to all engineering programmes at DTU and specialised courses to the mathematics, computer science, and other programmes. We offer continuing education courses and scientific advice within our research disciplines, and provide a portfolio of innovation activities for students and employees.

DTU is a technical university providing internationally leading research, education, innovation and scientific advice. Our staff of 6,000 advance science and technology to create innovative solutions that meet the demands of society, and our 11,200 students are being educated to address the technological challenges of the future. DTU is an independent university collaborating globally with business, industry, government and public agencies.