Prochaine session deeptech Skrub like a pro: clean, prepare and transform your data faster le 23 février – NOUVEAU

Prochaine session executive education Initiation à l’informatique quantique le 25 février – NOUVEAU

Prochaine session état de l’art La cybersécurité des systèmes de contrôle industriel le 10 mars

Prochaine session état de l’art Convergence calcul haute performance et intelligence artificielle le 10 mars – NOUVEAU

Prochaine session deeptech Mapping de données existantes vers des graphes de connaissances RDF le 13 mars – NOUVEAU

Toutes nos formations

Machine Learning and Cybersecurity

 Module état de l'art 
With fast development of AI techniques, it becomes a must to understand how AI would help predict future cyber security incidents and suggest proactive defence actions to mitigate potential cyber attacks. © Inria / Photo C. Morel

Session:

Aucune session disponible actuellement.

Contactez-nous !

Objectifs

We aim to reach two goals in this training program. First of all, we will introduce the popular AI-driven cyber security applications. We will especially focus on malware classification and malware clustering tasks that are necessary for most of commercialised Anti-Virus vendors. Second, we will propose several use cases of using AI toolboxes, such as Scikit-learn APIs, to analyse the statistics of malware samples, extract malware features and perform classification / clustering test. We will learn to how to set up a fair benchmark to evaluate the performances of AI-based security incident detection and showcase the potential impacts of dataset design over the evaluation result.

Target audience: R&D engineers and researchers.

Keywords: scikit-learn, malware classification, AI for security.

Pré-requis

  • Preliminary knowledge about scikit-learn, Numpy , Scipy packages in Python.

Programme

AI is beyond simply recognising images or videos, but also focusing on understand attacks, encoding knowledges learnt from past security incidents and recommend possible mitigation plans. We will cover an introduction to the potential use of AI for improving cyber safety service at first and then demonstrate the basic practices of AI algorithms to reach a data-driven security incident classification.

More specifically, we will include:

  • An introduction to the modern AI technologies and landed use cases of AI in the world of cyber security vendors,
  • A use case explanation about how to set up an AI pipeline for produce the summary of malware statistics, extraction of malware features and performance evaluation of malware classification result.
  • We explain the factors that may bring impacts over the malware classification results.

Intervenant(s)

  • Yufei Han

    Chargé de recherche Inria

    Dr Yufei Han is a researcher at Inria CIDRE project-team. He has been devoted himself into the research of adversarial AI techniques and AI-boosted cybersecurity applications for over 8 years. Before he joined Inria, he used to be senior principal researcher at Symantec Research Labs and post-doctoral researcher at Inria. He has published over 30 research papers on top-tiered research venues of AI and cyber security. He has also obtained 15 US patents on AI-based malware classification and intrusion detection systems.

Les prochaines sessions

2 jours

For more information

About prices

  • Price: €1 600 per participant
  • Discounted rates for groups of 5 or more (-10% for 5 to 9 participants, -20% for 10 or more participants).
  • Private training: This training course can be privatized from three participants, based on a flat rate for five people.
  • Duration: 2 days (9 a.m. to 12 p.m. and 2 p.m. to 5 p.m.);
  • Location: remote
  • Number of participants: up to 12 people;
  • Language: the training can be delivered in English;
  • Teaching methods: the training includes theoretical elements, some technical, others more general, to be taken into account in the strategy.
    The necessary materials and resources will be provided to encourage independent learning after the course.
  • Training assessment methods: At the end of the training, questionnaires will be sent out to assess the level of skills acquired.
  • Access times: The schedule is subject to the availability of Inria scientists. Early registration is recommended. Confirmation of the session will be given at least two weeks in advance.

Découvrir

Demander plus d'informations