Machine-Learning-as-a-Service (MLaaS) has become a widespread paradigm, making even the most complex Machine Learning models available for clients via, e.g., a pay-per-query principle. This allows users to avoid time-consuming processes of data collection, hyperparameter tuning, and model training. However, by giving their customers access to the (predictions of their) models, MLaaS providers endanger their intellectual property such as sensitive training data, optimised hyperparameters, or learned model parameters. In some cases, adversaries can create a copy of the model with (almost) identical behaviour using the the prediction labels only. While many variants of this attack have been described, only scattered defence strategies that address isolated threats have been proposed. To arrive at a comprehensive understanding why these attacks are successful and how they could be holistically defended against, a thorough systematisation of the field of model stealing is necessary. We address this by categorising and comparing model stealing attacks, assessing their performance, and exploring corresponding defence techniques in different settings. We propose a taxonomy for attack and defence approaches and provide guidelines on how to select the right attack or defence strategy based on the goal and available resources. Finally, we analyse which defences are rendered less effective by current attack strategies.

Evaluations of intrusion detection systems (IDS) require log datasets collected in realistic system environments. Existing testbeds therefore offer user simulations and attack scenarios that target specific use-cases. However, not only does the preparation of such testbeds require domain knowledge and time-consuming work, but also maintenance and modifications for other use-cases involve high manual efforts and repeated execution of tasks. In this article, we therefore propose to generate testbeds for IDS evaluation using strategies from model-driven engineering. In particular, our approach models system infrastructure, simulated normal behavior, and attack scenarios as testbed-independent modules. A transformation engine then automatically generates arbitrary numbers of testbeds, each with a particular set of characteristics and capable of running in parallel. Our approach greatly improves configurability and flexibility of testbeds and allows to reuse components across multiple scenarios. We use our proof-of-concept implementation to generate a labeled dataset for IDS evaluation that is published with this article.

The main goals and challenges for the life science communities in the Open Science framework are to increase reuse and sustainability of data resources, software tools, and workflows, especially in large‐scale data‐driven research and computational analyses. Here, we present key findings, procedures, effective measures and recommendations for generating and establishing sustainable life science resources based on the collaborative, cross‐disciplinary work done within the EOSC‐Life (European Open Science Cloud for Life Sciences) consortium. Bringing together 13 European life science research infrastructures, it has laid the foundation for an open, digital space to support biological and medical research. Using lessons learned from 27 selected projects, we describe the organisational, technical, financial and legal/ethical challenges that represent the main barriers to sustainability in the life sciences. We show how EOSC‐Life provides a model for sustainable data management according to FAIR (findability, accessibility, interoperability, and reusability) principles, including solutions for sensitive‐ and industry‐related resources, by means of cross‐disciplinary training and best practices sharing. Finally, we illustrate how data harmonisation and collaborative work facilitate interoperability of tools, data, solutions and lead to a better understanding of concepts, semantics and functionalities in the life sciences.

We introduce an algorithm that analyzes audio signals to extract chord-sequence information. The main goal of our approach lies in incorporating music theoretical knowledge without restricting the input data to a narrow range of musical styles. At the basis of our approach lies pitch detection using enhanced autocorrelation, supported by key detection and beat tracking. The chords themselves are identified by comparing generated and reference pitch class profiles. A smoothing algorithm is applied to the chord sequence which optimizes the number of chord changes and thus takes into consideration the comparatively stable nature of chords. In this paper we present an evaluation performed on a large set of 35 pieces of diverse music showing an average performance of 65% accuracy.