Real world data often have a long-tailed and open-ended distribution. A practical recognition system must classify among majority and minority classes, generalize from a few known instances, and acknowledge novelty upon a never seen instance. We define Open Long-Tailed Recognition (OLTR) as learning from such naturally distributed data and optimizing the classification accuracy over a balanced test set which include head, tail, and open classes.OLTR must handle imbalanced classification, few-shot learning, and open-set recognition in one integrated algorithm, whereas existing classification approaches focus only on one aspect and deliver poorly over the entire class spectrum. The key challenges are how to share visual knowledge between head and tail classes and how to reduce confusion between tail and open classes.We develop an integrated OLTR algorithm that maps an image to a feature space such that visual concepts can easily relate to each other based on a learned metric that respects the closed-world classification while acknowledging the novelty of the open world. Our so-called dynamic metaembedding combines a direct image feature and an associated memory feature, with the feature norm indicating the familiarity to known classes. On three large-scale OLTR datasets we curate from object-centric ImageNet, scenecentric Places, and face-centric MS1M data, our method consistently outperforms the state-of-the-art. Our code, datasets, and models enable future OLTR research and are publicly available at https://liuziwei7.github. io/projects/LongTail.html.

Even if people live in an arid desert, they know that plenty of water exists in the air they breathe. However, the reality tells us the atmospheric water cannot help to slake the world's thirst. Thus an important question occurs: what are the fundamental limits of atmospheric water harvesting that can be achieved in typical arid and semi-arid areas? Here, through a thorough review on the present advances of atmospheric water-harvesting technologies, we identify the achievements that have been acquired and evaluate the challenges and barriers that retard their applications. Lastly, we clarify our perspectives on how to search for a simple, scalable, yet cost-effective way to produce atmospheric water for the community and forecast the application of atmospheric water harvesting in evaporative cooling, such as electronic cooling, power plant cooling, and passive building cooling.

Objective: This meta-analysis aimed to estimate the global lifetime and 12-month prevalence of suicidal behavior, deliberate self-harm and non-suicidal self-injury in children and adolescents. Methods: A systematic search for relevant articles published between 1989 to 2018 was performed in multiple electronic databases. The aggregate 12-month and lifetime prevalence of suicidal behavior, deliberate self-harm, and non-suicidal self-injury were calculated based on the random-effects model. Subgroup analyses were performed to compare the prevalence according to school attendance and geographical regions. Results: A total of 686,672 children and adolescents were included. The aggregate lifetime and 12-month prevalence of suicide attempts was 6% (95% CI: 4.7–7.7%) and 4.5% (95% CI: 3.4–5.9%) respectively. The aggregate lifetime and 12-month prevalence of suicidal plan was 9.9% (95% CI: 5.5–17%) and 7.5% (95% CI: 4.5–12.1%) respectively. The aggregate lifetime and 12-month prevalence of suicidal ideation was 18% (95% CI: 14.2–22.7%) and 14.2% (95% CI: 11.6–17.3%) respectively. The aggregate lifetime and 12-month prevalence of non-suicidal self-injury was 22.1% (95% CI: 16.9–28.4%) and 19.5% (95% CI: 13.3–27.6%) respectively. The aggregate lifetime and 12-month prevalence of deliberate self-harm was 13.7% (95% CI: 11.0–17.0%) and 14.2% (95% CI: 10.1–19.5%) respectively. Subgroup analyses showed that full-time school attendance, non-Western countries, low and middle-income countries, and geographical locations might contribute to the higher aggregate prevalence of suicidal behaviors, deliberate self-harm, and non-suicidal self-injury. Conclusions: This meta-analysis found that non-suicidal self-injury, suicidal ideation, and deliberate self-harm were the three most common suicidal and self-harm behaviors in children and adolescents.

No abstract