Purpose The COVID-19 pandemic has many potential impacts on people with mental health conditions and on mental health care, including direct consequences of infection, effects of infection control measures and subsequent societal changes. We aimed to map early impacts of the pandemic on people with pre-existing mental health conditions and services they use, and to identify individual and service-level strategies adopted to manage these. Methods We searched for relevant material in the public domain published before 30 April 2020, including papers in scientific and professional journals, published first person accounts, media articles, and publications by governments, charities and professional associations. Search languages were English, French, German, Italian, Spanish, and Mandarin Chinese. Relevant content was retrieved and summarised via a rapid qualitative framework synthesis approach. Results We found 872 eligible sources from 28 countries. Most documented observations and experiences rather than reporting research data. We found many reports of deteriorations in symptoms, and of impacts of loneliness and social isolation and of lack of access to services and resources, but sometimes also of resilience, effective self-management and peer support. Immediate service challenges related to controlling infection, especially in inpatient and residential settings, and establishing remote working, especially in the community. We summarise reports of swiftly implemented adaptations and innovations, but also of pressing ethical challenges and concerns for the future. Conclusion Our analysis captures the range of stakeholder perspectives and experiences publicly reported in the early stages of the COVID-19 pandemic in several countries. We identify potential foci for service planning and research.

Deep neural networks (DNN) have been shown to be useful in a wide range of applications. However, they are also known to be vulnerable to adversarial samples. By transforming a normal sample with some carefully crafted human imperceptible perturbations, even highly accurate DNN make wrong decisions. Multiple defense mechanisms have been proposed which aim to hinder the generation of such adversarial samples. However, a recent work show that most of them are ineffective. In this work, we propose an alternative approach to detect adversarial samples at runtime. Our main observation is that adversarial samples are much more sensitive than normal samples if we impose random mutations on the DNN. We thus first propose a measure of 'sensitivity' and show empirically that normal samples and adversarial samples have distinguishable sensitivity. We then integrate statistical hypothesis testing and model mutation testing to check whether an input sample is likely to be normal or adversarial at runtime by measuring its sensitivity. We evaluated our approach on the MNIST and CIFAR10 datasets. The results show that our approach detects adversarial samples generated by state-of-the-art attacking methods efficiently and accurately.

No abstract

No abstract