Search

Een ontwerpgericht onderzoek naar de veiligheidsbeleving van meiden in de openbare ruimte

In recent years, there has been a great deal of studies about the optimisation of generating adversarial examples for Deep Neural Networks (DNNs) in a black-box environment. The use of gradient-based techniques to get the adversarial images in a minimal amount of input-output correspondence with the attacked model has been extensively studied. However, existing studies have not been discussing the effect of different gradient estimation techniques coherently. In this paper, a new one-point residual estimate is compared to the known two-point estimates. The findings in this paper show that the one-point residual estimate is not a viable option to decrease the number of queries to the attacked model. The accuracy of the attacks with the use of an one-point residual estimate maintains the same for weaker models. For stronger models, there is a slight decrease in accuracy at identical distortion levels. All estimates are tested on different PGD attacks on the MNIST and F-MNIST datasets using a 3-layer convolutional network., CSE3000 Research Project, Computer Science and Engineering

BACKGROUND: For some patients with advanced cancer not knowing prognosis is essential. Yet, in an era of informed decision-making, the potential protective function of unawareness is easily overlooked. We aimed to investigate 1) the proportion of advanced cancer patients preferring not to know prognosis; 2) the reasons underlying patients' prognostic information preference; 3) the characteristics associated with patients' prognostic information preference; and 4) the concordance between physicians' perceived and patients' actual prognostic information preference. METHODS: This is a cross-sectional study with structured surveys (PROSPECT). Medical and thoracic oncologists included patients (n = 524), from seven Dutch hospitals, with metastatic/inoperable cancer and an expected median overall survival of ≤ 12 months. For analysis, descriptive statistics and logistic regression models were used. RESULTS: Twenty-five to 31% of patients preferred not to know a general life expectancy estimate or the 5/2/1-year mortality risk. Compared to patients preferring to know prognosis, patients preferring unawareness more often reported optimism, avoidance and inability to comprehend information as reasons for wanting limited information; and less often reported expectations of others, anxiety, autonomy and a sense of control as reasons for wanting complete information. Females (p < .05), patients receiving a further line of systemic treatment (p < .01) and patients with strong fighting spirit (p < .001) were more likely to prefer not to know prognosis. Concordance between physicians' perceived and patients' actual prognostic information preference was poor (kappa = 0.07). CONCLUSIONS: We encourage physicians to explore patients' prognostic information preferences and the underlying reasons explicitly, enabling individually tailored communication. Future studies may investigate changes in patients' prognostic information prefer

Block signalling and Automatic Train Protection are two of the main railway safety systems to control the risk of train accidents. The combination of the existing Dutch block signalling system NS’54 and ATP-system ATB-EG is functioning well, but has some drawbacks. The systems have been designed over 60 years ago with proven technology of that time. Components are old and will have to be replaced in the near future. The safety functionality of the legacy systems is limited. Moreover, the capacity of the Dutch railway network with NS’54/ATB-EG is almost fully used while demand grows. Solutions are required to match the demand growth. The European Rail Traffic Managements System (ERTMS) is proposed to be the new standard European safety system. Train control is included in the European Train Control System (ETCS). It offers benefits in the form of interoperability, increased safety, speed, capacity and/or reliability over the national system. ERTMS/ETCS Hybrid Level 3 is an integrated cab-signalling and train protection system that combines train position information, onboard train integrity confirmation and trackside train detection. Compared to the legacy system NS'54/ATB-EG and the proposed system ERTMS/ETCS Level 2 it offers more railway capacity and allows for a reduction in trackside train detection. This thesis shows that the infrastructure occupation can drop from 84% to 66,7% when implementing ERTMS/ETCS Hybrid Level 3. Reducing the amount of trackside train detection to the minimum and transitioning from track circuits to axle counters can result in a decrease of track unavailability related to train protection systems of over 40% compared to the existing configuration., Civil Engineering

Migration studies have developed rapidly as a research field over the past decades. This article provides an empirical analysis not only on the development in volume and the internationalization of the field, but also on the development in terms of topical focus within migration studies over the past three decades. To capture volume, internationalisation, and topic focus, our analysis involves a computer-based topic modelling of the landscape of migration studies. Rather than a linear growth path towards an increasingly diversified and fragmented field, as suggested in the literature, this reveals a more complex path of coming of age of migration studies. Although there seems to be even an accelerated growth for migration studies in terms of volume, its internationalisation proceeds only slowly. Furthermore, our analysis shows that rather than a growth of diversification of topics within migration topic, we see a shift between various topics within the field. Finally, our study shows that there is no consistent trend to more fragmentation in the field; in contrast, it reveals a recent recovery of connectedness between the topics in the field, suggesting an institutionalisation or even theoretical and conceptual coming of age of migration studies.