On 22 January 2020, the first anniversary of the Aachen Treaty, the French National Institute for Information Technology, Inria, and DFKI signed a Memorandum of Understanding. The two scientific institutions agreed to strengthen their cooperation in the field of AI significantly, to structure and formalize their long-standing scientific collaboration.
Inria and DFKI will work together in a joint strategic research and innovation agenda in the areas of AI for Industry 4.0, AI infrastructures, AI and cybersecurity, human-robot collaboration, wearables, and other topics. A core part of the agreement consists of the strong shared commitment to the European AI initiative CLAIRE (Confederation of Labs for AI Research in Europe). CLAIRE aims to bring European AI researchers closer together and collectively advance European research for AI that benefits people while respecting fundamental European values.
Concrete measures include the implementation of joint research and innovation projects, which partly build on both organizations' existing projects. Furthermore, new topics will be defined and elaborated in joint workshops. The first projects have been launched in the pilot year 2020.
The fulfilment of individual customer requirements is becoming an increasingly decisive factor when considering the competitiveness of companies. The make-to-order or engineer-to-order production resulting from this not only have an impact on the manufacturing process, but also on the entire value chain of the product - starting with the raw material supplier and ending with the end customer. Compared to mass and serial production, this dynamic poses challenges for the supply chain - a continuous adaptation to requirements throughout the entire supply chain. This is no longer manageable with static optimization methods. Using real-time data and methods of Artificial Intelligence, an intelligent system is to be developed that enables a proactive and semi-automated adaptation and optimization of individual manufacturing processes, taking into account current and predicted external and internal business events and situations.
Virtually all NLP systems nowadays use vector representations of words, a.k.a. word embeddings. Similarly, the processing of language combined with vision or other sensory modalities employs multimodal embeddings. While embeddings do embody some form of semantic relatedness, the exact nature of the latter remains unclear. This loss of precise semantic information can affect downstream tasks. The goals of IMPRESS are to investigate the integration of semantic and common sense knowledge into linguistic and multimodal embeddings and the impact on selected downstream tasks. IMPRESS will also develop open source software and lexical resources, focusing on video activity recognition as a practical testbed.
MePheSTO is an interdisciplinary research project that aims to develop a scientifically grounded, artificial intelligence based methodology for identifying and classifying measurable, and thus objective, digital phenotypes of psychiatric disorders. The aim of the project is to develop a technological platform for the cientifically validation of phenotypes for psychiatric disorders based on multimodal inputs such as speech, video and biosignals from clinical social interactions. For this purpose, researchers collect data from video recordings, conversations, but also from brain or heart activity (EEG, ECG).
The aim of the MOVEON project is to develop a novel generation of visual positioning systems that goes beyond classical localization and mapping, which focuses currently only on point cloud reconstruction. In contrast, our aim is to allow for 6DoF positioning and global scene understanding in wild and dynamic environments (e.g. crowded streets) that scales up nicely with the size of the environment, and that can be used persistently over time by reusing consistent maps. MOVEON will push forward the state of the art in vision-based, spatio-temporal scene understanding by merging novel machine-learning approaches with geometrical reasoning. Deep-learning-based recognition and understanding of high-level concepts such as vanishing points or large object classes will serve as unitary building blocks for a spatio-temporal localization and environment reconstruction that will use geometric reasoning as underlying support.