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
Universities have become more engaged or entrepreneurial, forging deeper relations with society beyond the economic sphere. To foster, structure, and institutionalize a broader spectrum of engagement, new types of intermediary organizations are created, going beyond the “standard” technology transfer oces, incubators, and science parks. This paper conceptualizes the role of such new-style intermediaries as facilitator, enabler, and co-shaper of university–society interaction, making a distinction between the roles of facilitation, configuration, and brokering. As a case study, the paper presents the Knowledge Mile in Amsterdam as a novel form of hyper local engagement of a university with its urban surroundings that connects the challenges of companies and organisations in the street to a broad range of educational and research activities of the university, as well as to rebrand the street.
Calls for transformative change and participatory modes of knowledge production demand researchers to assume new roles. This paper synthesizes the literature on knowledge co-production and researcher roles to explore challenges for researchers involved in transdisciplinary environmental management projects. Our research methods combine a scoping review and reflections on personal experiences with three transdisciplinary projects. To conceptualize researcher roles in transdisciplinary knowledge co-production, we distinguish between three spaces: knowledge, formal policy, and stakeholder. Knowledge co-production requires collaboration between actors from different spaces and integration of diverse knowledge sources and types. Depending on whether researchers adopt knowledge-oriented, change-oriented or intermediating roles, they will experience different challenges. When researchers combine knowledge development with change-oriented and/or intermediating roles, they encounter new challenges, such as, maintaining independence or objectivity. To assist researchers in transdisciplinary projects, we conclude with a checklist of four elements to reflect upon: orientation, norms and values, expectations and resources.
LINK
INXCES will use and enhance innovative 3D terrain analysis and visualization technology coupled with state-of-the-art satellite remote sensing to develop cost-effective risk assessment tools for urban flooding, aquifer recharge, ground stability and subsidence. INXCES will develop quick scan tools that will help decision makers and other actors to improve the understanding of urban and peri-urban terrains and identify options for cost effective implementation of water management solutions that reduce the negative impacts of extreme events, maximize beneficial uses of rainwater and stormwater for small to intermediate events and provide long-term resilience in light of future climate changes. The INXCES approach optimizes the multiple benefits of urban ecosystems, thereby stimulating widespread implementation of nature-based solutions on the urban catchment scale.INXCES will develop new innovative technological methods for risk assessment and mitigation of extreme hydroclimatic events and optimization of urban water-dependent ecosystem services at the catchment level, for a spectrum of rainfall events. It is widely acknowledged that extreme events such as floods and droughts are an increasing challenge, particularly in urban areas. The frequency and intensity of floods and droughts pose challenges for economic and social development, negatively affecting the quality of life of urban populations. Prevention and mitigation of the consequences of hydroclimatic extreme events are dependent on the time scale. Floods are typically a consequence of intense rainfall events with short duration. In relation to prolonged droughts however, a much slower timescale needs to be considered, connected to groundwater level reductions, desiccation and negative consequences for growing conditions and potential ground – and building stability.INXCES will take a holistic spatial and temporal approach to the urban water balance at a catchment scale and perform technical-scientific research to assess, mitigate and build resilience in cities against extreme hydroclimatic events with nature-based solutions.INXCES will use and enhance innovative 3D terrain analysis and visualization technology coupled with state-of-the-art satellite remote sensing to develop cost-effective risk assessment tools for urban flooding, aquifer recharge, ground stability and subsidence. INXCES will develop quick scan tools that will help decision makers and other actors to improve the understanding of urban and peri-urban terrains and identify options for cost effective implementation of water management solutions that reduce the negative impacts of extreme events, maximize beneficial uses of rainwater and stormwater for small to intermediate events and provide long-term resilience in light of future climate changes. The INXCES approach optimizes the multiple benefits of urban ecosystems, thereby stimulating widespread implementation of nature-based solutions on the urban catchment scale.