Call 2014 Topics
Note: the short topic descriptions below are provided in view of the conference. The full topic descriptions, to appear in the announcement of the Call 2014, will result from the conference.
1. Resilient Trustworthy Cyber-Physical Systems
Cyber-physical systems (CPS) refer to novel hardware and software compositions creating smart, autonomously acting devices, enabling efficient end-to-end workflows and new forms of user-machine interaction. In manifold emerging application domains such as health care, traffic management or energy supply CPS carry a high potential for creating new markets and solutions to societal hazards, but impose highest requirements to quality in terms of resilience, safety, security and privacy. However, the heterogeneous, evolving and distributed nature of CPS bears major challenges to continuously assure these quality requirements employing state of the art methods and technologies. Foundational research efforts are needed to achieve a predictable quality level in an efficient, traceable and measurable way, coping efficiently with external and internal changes, supporting necessary transitions between mechanical, electrical and software engineering, as well as integrating management, design and deployment aspects.
- Security and safety requirements engineering
- Security infrastructures
- Intrusion detection
- Risk management
- Quality and risk models
- Model-based software and systems engineering
- Multi-level architectures
- Verification and validation
- Automated Testing
- Security testing
- Security and privacy protocols
- Certification processes
2. Human Language Understanding
Having a machine understand language like a human being can be considered as the epitome of Artificial Intelligence, as exemplified by the design of the Turing test. Even though the domain of automatic language processing has made steady progress over the last decades and some applications are out on the market, the machine is still far from reaching human performance. New approaches are needed, in particular to model high-level, semantic and pragmatic knowledge in a robust fashion. For that purpose, the machine learning approach which has proved very successful to train models from linguistic data should be extended to use more varied data, potentially covering the whole gamut of stimuli a human can learn from, in a much more multidisciplinary approach.
- Machine Reading, Question Answering, Dialog Systems, Automatic Speech Recognition, Optical Character and Handwriting Recognition, Opinion and Sentiment Analysis
- Machine Learning, Semi-Supervised Learning, Reinforcement Learning, Active Learning
- Semantic Analysis, Context Modelling, Concept Learning, Deep Learning
- Embodied intelligence, multimodal data processing, information fusion