We present a novel architecture for an AI system that allows a priori knowledge to combine with deep learning. In traditional neural networks, all available data is pooled at the input layer. Our alternative neural network is constructed so that partial representations (invariants) are learned in the intermediate layers, which can then be combined with a priori knowledge or with other predictive analyses of the same data. This leads to smaller training datasets due to more efficient learning. In addition, because this architecture allows inclusion of a priori knowledge and interpretable predictive models, the interpretability of the entire system increases while the data can still be used in a black box neural network. Our system makes use of networks of neurons rather than single neurons to enable the representation of approximations (invariants) of the output.
LINK
The authors consider the reality that endless economic growth on a finite planet is unsustainable, especially if society has exceeded ecological limits. The paper examines various aspects of society's endless growth predicament. It reviews the idea that there are 'limits to growth'; it then considers the 'endless growth mantra' within society. The paper then considers the 'decoupling' strategy and its merits, and argues that it is, at best, a partial solution to the problem. The key social problem of denial of our predicament is considered, along with the contribution of anthropocentric modernism as a worldview that aids and abets that denial. Finally, the paper outlines some potential solutions to our growth predicament. https://www.ecologicalcitizen.net/article.php?t=insanity-endless-growth https://www.linkedin.com/in/helenkopnina/
MULTIFILE
De Digitale Universiteit (DU) performed a quickscan to determine the usability of the IMS Question and Test Interoperability (QTI) specification as a format to store questions and tests developed for and by the consortium. The original report is available in Dutch from the website of De Digitale Universiteit. This is an unofficial translation in English of that report.
To reach the European Green Deal by 2050, the target for the road transport sector is set at 30% less CO2 emissions by 2030. Given the fact that heavy-duty commercial vehicles throughout Europe are driven nowadays almost exclusively on fossil fuels it is obvious that transition towards reduced emission targets needs to happen seamlessly by hybridization of the existing fleet, with a continuously increasing share of Zero Emission vehicle units. At present, trailing units such as semitrailers do not possess any form of powertrain, being a missed opportunity. By introduction of electrically driven axles into these units the fuel consumption as well as amount of emissions may be reduced substantially while part of the propulsion forces is being supplied on emission-free basis. Furthermore, the electrification of trailing units enables partial recuperation of kinetic energy while braking. Nevertheless, a number of challenges still exist preventing swift integration of these vehicles to daily operation. One of the dominating ones is the intelligent control of the e-axle so it delivers right amount of propulsion/braking power at the right time without receiving detailed information from the towing vehicle (such as e.g. driver control, engine speed, engine torque, or brake pressure, …etc.). This is required mainly to ensure interoperability of e-Trailers in the fleets, which is a must in the logistics nowadays. Therefore the main mission of CHANGE is to generate a chain of knowledge in developing and implementing data driven AI-based applications enabling SMEs of the Dutch trailer industry to contribute to seamless energetic transition towards zero emission road freight transport. In specific, CHANGE will employ e-Trailers (trailers with electrically driven axle(s) enabling energy recuperation) connected to conventional hauling units as well as trailers for high volume and extreme payload as focal platforms (demonstrators) for deployment of these applications.
The IMPULS-2020 project DIGIREAL (BUas, 2021) aims to significantly strengthen BUAS’ Research and Development (R&D) on Digital Realities for the benefit of innovation in our sectoral industries. The project will furthermore help BUas to position itself in the emerging innovation ecosystems on Human Interaction, AI and Interactive Technologies. The pandemic has had a tremendous negative impact on BUas industrial sectors of research: Tourism, Leisure and Events, Hospitality and Facility, Built Environment and Logistics. Our partner industries are in great need of innovative responses to the crises. Data, AI combined with Interactive and Immersive Technologies (Games, VR/AR) can provide a partial solution, in line with the key-enabling technologies of the Smart Industry agenda. DIGIREAL builds upon our well-established expertise and capacity in entertainment and serious games and digital media (VR/AR). It furthermore strengthens our initial plans to venture into Data and Applied AI. Digital Realities offer great opportunities for sectoral industry research and innovation, such as experience measurement in Leisure and Hospitality, data-driven decision-making for (sustainable) tourism, geo-data simulations for Logistics and Digital Twins for Spatial Planning. Although BUas already has successful R&D projects in these areas, the synergy can and should significantly be improved. We propose a coherent one-year Impuls funded package to develop (in 2021): 1. A multi-year R&D program on Digital Realities, that leads to, 2. Strategic R&D proposals, in particular a SPRONG/sleuteltechnologie proposal; 3. Partnerships in the regional and national innovation ecosystem, in particular Mind Labs and Data Development Lab (DDL); 4. A shared Digital Realities Lab infrastructure, in particular hardware/software/peopleware for Augmented and Mixed Reality; 5. Leadership, support and operational capacity to achieve and support the above. The proposal presents a work program and management structure, with external partners in an advisory role.
Façades have a high environmental and economic impact: they contribute 10-30% to GHG emissions and 30-40% of the building investment of new buildings [1]. Modern façades are highly optimized complex systems that consist of multiple components with varying life cycles [2]; however, many of the materials they employ are critical, and have a high CO2 footprint [3, 4]. New bio-composite facades products have emerged (a) whose mechanical properties are comparable to those of aluminum or glass fibre; (b) have a lower energy footprint; and (c) can fully or partially biodegrade [5]. Moreover, primary material sourcing from different waste streams can significantly lower the end products’ pricing. Still, their aesthetic qualities have not been sufficiently explored, so the scalability of their production remains limited. This project will develop specific combinations of bio-composites using food waste fillers and a biopolymer resin. Sheet samples will be made from these combinations and further tested against their mechanical properties, water resistance, aging and weathering. A Life Cycle Analysis will further consolidate the samples’ energy footprint. A new facade cladding tile product system with complex geometry using the overall best performing material composition will be designed and prototyped [17]. Emphasis will be given to the aesthetical properties of the tiles and their demountability. The system tiles will be further applied and tested at 1:1 scale, at The Green Village. During the project, an advisory board consisting of several companies within the building industry will be systematically consulted and their feedback will help the overall design process and their respective end products.
