The adoption of machine learning and deep learning is on the rise in the cybersecurity domain where these AI methods help strengthen traditional system monitoring and threat detection solutions. However, adversaries too are becoming more effective in concealing malicious behavior amongst large amounts of benign behavior data. To address the increasing time-to-detection of these stealthy attacks, interconnected and federated learning systems can improve the detection of malicious behavior by joining forces and pooling together monitoring data. The major challenge that we address in this work is that in a federated learning setup, an adversary has many more opportunities to poison one of the local machine learning models with malicious training samples, thereby influencing the outcome of the federated learning and evading detection. We present a solution where contributing parties in federated learning can be held accountable and have their model updates audited. We describe a permissioned blockchain-based federated learning method where incremental updates to an anomaly detection machine learning model are chained together on the distributed ledger. By integrating federated learning with blockchain technology, our solution supports the auditing of machine learning models without the necessity to centralize the training data. Experiments with a realistic intrusion detection use case and an autoencoder for anomaly detection illustrate that the increased complexity caused by blockchain technology has a limited performance impact on the federated learning, varying between 5 and 15%, while providing full transparency over the distributed training process of the neural network. Furthermore, our blockchain-based federated learning solution can be generalized and applied to more sophisticated neural network architectures and other use cases.

Strongly rooted in the Internet of Things and Cyber-Physical Systems-enabled manufacturing, disruptive paradigms like the Factory of the Future and Industry 4.0 envision knowledge-intensive industrial intelligent environments where smart personalized

Cyber threat intelligence (CTI) sharing is the collaborative effort of sharing information about cyber attacks to help organizations gain a better understanding of threats and proactively defend their systems and networks from cyber attacks. The challenge that we address is the fact that traditional indicators of compromise (IoC) may not always capture the breath or essence of a cyber security threat or attack campaign, possibly leading to false alert fatigue and missed detections with security analysts. To tackle this concern, we designed and evaluated a CTI solution that complements the attribute and tagging based sharing of indicators of compromise with machine learning (ML) models for collaborative threat detection. We implemented our solution on top of MISP, TheHive, and Cortex—three state-of-practice open source CTI sharing and incident response platforms—to incrementally improve the accuracy of these ML models, i.e., reduce the false positives and false negatives with shared counter-evidence, as well as ascertain the robustness of these models against ML attacks. However, the ML models can be attacked as well by adversaries that aim to evade detection. To protect the models and to maintain confidentiality and trust in the shared threat intelligence, we extend our previous research to offer fine-grained access to CP-ABE encrypted machine learning models and related artifacts to authorized parties. Our evaluation demonstrates the practical feasibility of the ML model based threat intelligence sharing, including the ability of accounting for indicators of adversarial ML threats.

The application of artificial intelligence enhances the ability of sensor and networking technologies to realize smart systems that sense, monitor and automatically control our everyday environments. Intelligent systems and applications often automate decisions based on the outcome of certain machine learning models. They collaborate at an ever increasing scale, ranging from smart homes and smart factories to smart cities. The best performing machine learning model, its architecture and parameters for a given task are ideally automatically determined through a hyperparameter tuning process. At the same time, edge computing is an emerging distributed computing paradigm that aims to bring computation and data storage closer to the location where they are needed to save network bandwidth or reduce the latency of requests. The challenge we address in this work is that hyperparameter tuning does not take into consideration resource trade-offs when selecting the best model for deployment in smart environments. The most accurate model might be prohibitively expensive to computationally evaluate on a resource constrained node at the edge of the network. We propose a multi-objective optimization solution to find acceptable trade-offs between model accuracy and resource consumption to enable the deployment of machine learning models in resource constrained smart environments. We demonstrate the feasibility of our approach by means of an anomaly detection use case. Additionally, we evaluate the extent that transfer learning techniques can be applied to reduce the amount of training required by reusing previous models, parameters and trade-off points from similar settings.

(1) Background: Container orchestration frameworks provide support for management of complex distributed applications. Different frameworks have emerged only recently, and they have been in constant evolution as new features are being introduced. This reality makes it difficult for practitioners and researchers to maintain a clear view of the technology space. (2) Methods: we present a descriptive feature comparison study of the three most prominent orchestration frameworks: Docker Swarm, Kubernetes, and Mesos, which can be combined with Marathon, Aurora or DC/OS. This study aims at (i) identifying the common and unique features of all frameworks, (ii) comparing these frameworks qualitatively and quantitatively with respect to genericity in terms of supported features, and (iii) investigating the maturity and stability of the frameworks as well as the pioneering nature of each framework by studying the historical evolution of the frameworks on GitHub. (3) Results: (i) we have identified 124 common features and 54 unique features that we divided into a taxonomy of 9 functional aspects and 27 functional sub-aspects. (ii) Kubernetes supports the highest number of accumulated common and unique features for all 9 functional aspects; however, no evidence has been found for significant differences in genericity with Docker Swarm and DC/OS. (iii) Very little feature deprecations have been found and 15 out of 27 sub-aspects have been identified as mature and stable. These are pioneered in descending order by Kubernetes, Mesos, and Marathon. (4) Conclusion: there is a broad and mature foundation that underpins all container orchestration frameworks. Likely areas for further evolution and innovation include system support for improved cluster security and container security, performance isolation of GPU, disk and network resources, and network plugin architectures.