To date, the design of ethical machine learning (ML) algorithms has been dominated by technology owners and remains broadly criticized for strategically seeking to avoid legally enforceable restrictions. In order to foster trust in ML technologies, society demands technology designers to deeply engage all relevant stakeholders in the ML development. This ERC project aims at responding to this call with a society-aware approach to ML (SAML). My goal is to enable the collaborative design of ML algorithms, so that they are not only driven by economic interests of the technology owners but are agreed upon by all stakeholders, and ultimately, trusted by society. To this end, I aim to develop multi-party ML algorithms that explicitly account for the goals of different stakeholders---i.e., owners, those experts that design the algorithm (e.g., technology companies); consumers, those that are affected by the algorithm (e.g., users); and regulators, those experts that set the regulatory framework for their use (e.g., policy makers). The proposed methodology will enable quantifying and jointly optimizing the business goals of the owners (e.g., profit); the benefits of the consumers (e.g., information access); and the risks defined by the regulators (e.g., societal polarization). The SAML project involves a high-risk/high-gain paradigm shift from an owner-centered to a society-centered (multi-party) ML design. On the one hand, it will require significant and challenging methodological innovations at every stage of the ML development: from the data collection all the way to the algorithm learning. On the other hand, it will impact how ML technologies are deployed in society by enabling an informed discussion among different stakeholders and, in general, by society about these new technologies. The results of this project will provide the urgently needed methodological foundations to ensure that these new technologies are at the service of society.

Today’s industry is more vulnerable to cyberattacks than ever. The biggest threat comes from advanced persistent threats that targets the sensitive data of a specific company. Such a threat may come along as an innocuous app that starts its malicious behavior only when the mobile logs into the corporate network. At the same time, such threats can be made undetectable through testing or code analysis. The ERC SPECMATE project has developed a technology named BOXMATE that protects against unexpected changes of app behavior and thus drastically reduces the attack surface of mobile applications. The key idea is to mine app behavior by executing generated tests, systematically exploring the program’s accesses to sensitive data. During production, the app then is placed in a sandbox, which prohibits accesses not seen during testing. This combination of test generation and sandboxing effectively protects against advanced persistent threats. To access sensitive data during production, the app already must do so during testing—where tracing makes it easy to discover and assess. BOXMATE neither does not need to collect user data: All app behavior is assessed during testing already. Finally, BOXMATE requires no knowledge about source or binary code, and thus easily handles arbitrarily obfuscated or obscure third-party apps. BOXMATE is currently being patented worldwide. We want to turn the BOXMATE approach into a full mobile security solution for corporate and end users. This proposal aims at producing a full-fledged prototype that can be demonstrated to potential customers, most notably app vendors and mobile infrastructure providers; as well as developing an adequate marketing strategy exploring and responding to the needs of the market. This proposal is fueled by the principal investigator, Andreas Zeller, one of the world’s leading experts in software test generation and specification mining.

The pivotal role of software in our modern world mandates strong requirements on quality, correctness, and reliability of software systems. In software development and maintenance, the ability to understand program artifacts plays a key role for programmers to fulfill these requirements. Despite significant progress, research on program comprehension has a fundamental limitation: program comprehension is a cognitive process that cannot be directly observed, which leaves considerable room for misinterpretation, uncertainty, and confounders. In Brains On Code, we will develop a neuroscientific foundation of program comprehension. Instead of merely observing whether there is a difference regarding program comprehension (e.g., between two programming methods), we aim at precisely and reliably determining the key factors that cause the difference. This is especially challenging as humans are the subjects of study, and inter-personal variance and other confounding factors obfuscate the results. The key idea of Brains On Code is to leverage established methods from cognitive neuroscience to obtain insights into the underlying processes and influential factors of program comprehension. Brains On Code will pursue a multimodal approach that integrates different neuro-physiological measures as well as a cognitive computational modeling approach to establish the theoretical foundation. This way, Brains On Code will lay the foundations of measuring and modeling program comprehension and offer substantial feedback for programming methodology, language design, and education. Addressing longstanding foundational questions such as "How can we reliably measure program comprehension?", "What makes a program hard to understand?", and "What skills should programmers have?" will become into reach. A success of Brains On Code would not only help answer these questions, but also provide an outline for applying the methodology beyond program code (models, specifications, etc.).

TALENTS aims to build up an international programme for interdisciplinary and transsectorial training of 15 doctoral candidates (DCs) as needed for the next generation of drug researchers. Research projects are structured in 3 clusters (PCs): PC1 will investigate the microbiome by transcriptomic and metagenomic analysis of diseases to identify and isolate microorganisms/metabolites for therapeutically modifying the microbiota. PC2 will focus on pathoblockers targeting virulence factors as alternative to traditional antibiotics with lower risk of resistance development. PC3 will address the delivery needs across biological barriers, to achieve efficient and targeted transport to the site of action. Throughout the PC structure of 15 individual, but interconnected research projects, we aim to develop and exchange disease models, analytical methods and microbiome-related results. By generating knowledge on correlations and causalities between microbiota and disease TALENTS will advance the development of novel microbiome-modulating therapies TALENTS is builds on an earlier joint venture of Saarland University, University Clinics Saarland and Helmholtz Institute for Pharmaceutical Research Saarland to foster transdisciplinary doctoral training between different faculties and institutions. In addition, TALENTS is intersectorial by implementing a unique supervision structure for each DC with an external expert advisor and translation-oriented training for individual career development, also including secondments in a complementary industrial or clinical sector. Working on challenging research projects, the candidates will acquire specific knowledge and skills at the interface of clinical medicine, microbiology and pharmaceutical science, pivotal for state-of-the-art infection research and microbiome interventions. By collaborations of experts from natural sciences, medicine, and bioinformatics, TALENTS provides high-level multidisciplinary and intersectorial training.

Generating huge amounts of visual data, be it images or videos, has never been easier than today. This creates a growing demand for lossy codecs (coders and decoders) that produce visually convincing results also for very high compression rates. Popular transform-based codecs such as JPEG and JPEG 2000 have reached a state where one cannot expect significant improvements anymore. To go beyond their limitations, fundamentally different ideas are needed. Inpainting-based codecs can change this situation. They store only a small, carefully optimised part of the data. In the decoding step, the missing information is filled in with a suitable inpainting mechanism. A successful realisation of inpainting-based codecs can offer decisive advantages over transform-based codecs: The stored information is more intuitive and closer to the mechanisms of human perception. Moreover, the concept is very flexible: It allows to integrate a number of different features and can be tailored towards dedicated applications. Most importantly, the higher the compression rate, the larger are the qualitative advantages over transform-based codecs. However, the potential of these codecs is widely unexplored so far, since difficult fundamental problems must be solved first. This includes optimisation of the data and the inpainting process, sophisticated data coding, and the design of real-time capable sequential and parallel numerical algorithms. We are committed to addressing all these challenges in an integrated approach: We cover the entire spectrum from its theoretical foundations over benchmarking and highly efficient numerical algorithms to codecs for specific applications, and a real-time 4K video player as demonstrator. This will lift inpainting methods from a visually pleasant image editing tool to a fundamental paradigm in coding. Research results that enter forthcoming coding standards will also have an impact on everybody's daily life.