A growing number of applications rely on AI-based solutions to carry-out mission-critical tasks, many of which are of temporal nature, dealing with ever-evolving flows of information. Crucial for mitigating threats and taking advantage of opportunities in such domains, is the ability to forecast imminent situations and critical complex events ahead of time. EVENFLOW will develop hybrid learning techniques for complex event forecasting, which combine deep learning with logic-based learning and reasoning into neuro-symbolic forecasting models. The envisioned methods will combine (i) neural representation learning techniques, capable of constructing event-based features from streams of perception-level data with (ii) powerful symbolic learning and reasoning tools, that utilize such features to synthesize high-level, interpretable patterns of critical situations to be forecast. Crucial in the EVENFLOW approach is the online nature of the learning methods, which makes them applicable to evolving data flows and allows to utilize rich domain knowledge that is becoming available progressively, over time. To deal with the brittleness of neural predictors and the high volume/velocity of temporal data flows, the EVENFLOW techniques will rely on novel, formal verification techniques for machine learning, in addition to a suite of scalability algorithms for federated training and incremental model construction. The learnt forecasters will be interpretable and scalable, allowing for fully explainable insights, delivered in a timely fashion and enabling proactive decision making. EVENFLOW will be evaluated on three challenging use cases related to oncological forecasting in precision medicine, safe and efficient behavior of autonomous transportation robots in smart factories and reliable life cycle assessment of critical infrastructure.

The amount of digital content available to creative industries is growing exponentially, driven by the ubiquitous use of smartphones and the proliferation of social media: (1) there is a tremendous increase in the amount of photographic content (more than 1.8 billion photos are uploaded to social media platforms each day); (2) the ongoing transformation of factual, entertainment and social media publishers and platforms from textual and photo-centric format to video-driven format (more than 400 hours of video are being uploaded to YouTube each minute); (3) the increasing impact of 3D and virtual reality for providing immersive storytelling experiences, offering new ways of audience engagement and monetization for content creators in the upcoming years. Acknowledging the above, the following critical questions become imminent in both content production and dissemination contexts: how to repurpose this massive amount of content; what kind of innovative tools are most suitable for this process; and finally how these tools can offer new forms of monetization possibilities for creative industries' professionals. FotoInMotion, sets to solve these critical questions and provide an innovative solution to the repurposing of content by offering automated tools for innovative contextual data extraction, object recognition, creative transformation, editing and text animation as well as state of the art 3D conversion options that allow content creators to transform their photos into highly engaging spatial and three dimensional video experiences. FotoInMotion will focus on three major creative industries sectors: photojournalism to develop interactive photo driven stories; fashion, by opening up new forms of marketing, product placement and event coverage; and festivals, by enabling PR and publicity managers to communicate the festival experiences and engage audiences through immersive communication and repurposing festival archives.

The main objective of the project is to design and implement a parameterized, knowledge-based, multi-target food sensitive mini-portable system, with heterogeneous micro-scale photonics for on-the-spot food quality sensing and shelf-life prediction. In particular, the miniaturized smart integrated system will be able to detect food hazards, spoilage (incl. early sign of spoilage) and food fraud through the combined bio-chemical data analysis and additionally will be able to perform food components/additives analysis, food identification and prediction of food shelf-life. The following use case will be addressed during the project: Use case 1: Detection of mycotoxins in various grains and nuts. Aflatoxins detection. A simple, convenient ultraviolet test makes it possible to detect the possible presence of aflatoxin. Use case 2: Detection of early sign of spoilage and spoilage in fruits, vegetables, meat, fish: combined with estimation on product expiration date. Use case 3: Detection of food fraud: Adulteration of alcoholic beverages, oil, milk and meat. 3 sensor devices will be integrated in the miniaturised smart sensor node: i) a MEMS-based near IR spectrometer (950-1900 nm), ii) a UV-VIS spectrometer (450-900 nm) and iii) a micro-camera. Moreover 3 light sources will also be integrated to support the sensing functionality: i) UV-LED, ii) white LED and iii) a miniaturised IR emitter. Smart signal processing of the spectrum images will be performed by an advanced microcontroller, integrated in the sensing device. The data will be communicated to a smartphone device, where the spectroscopy analysis will take place with the help of a cloud-base application connected to a reference database. Advanced detection algorithms will be deployed both in the level of cloud and the smartphone application. PhasmaFOOD system will enable common consumers for on the spot food quality sensing and shelf-life prediction.