Classic supervised learning makes the closed-world assumption that the classes seen in testing must have appeared in training. However, this assumption is o en violated in real-world applications. For example, in a social media site, new topics emerge constantly and in e-commerce, new categories of products appear daily. A model that cannot detect new/unseen topics or products is hard to function well in such open environments. A desirable model working in such environments must be able to (1) reject examples from unseen classes (not appeared in training) and (2) incrementally learn the new/unseen classes to expand the existing model. is is called open-world learning (OWL). is paper proposes a new OWL method based on meta-learning. e key novelty is that the model maintains only a dynamic set of seen classes that allows new classes to be added or deleted with no need for model re-training. Each class is represented by a small set of training examples. In testing, the meta-classi er only uses the examples of the maintained seen classes (including the newly added classes) on-the-y for classi cation and rejection. Experimental results with e-commerce product classi cation show that the proposed method is highly e ective 1 .

Background Prenatal adverse environments can cause fetal intrauterine growth retardation (IUGR) and higher susceptibility to multiple diseases after birth, related to multi-organ development programming changes mediated by intrauterine overexposure to maternal glucocorticoids. As a glucocorticoid barrier, P-glycoprotein (P-gp) is highly expressed in placental syncytiotrophoblasts; however, the effect of P-gp on the occurrence of IUGR remains unclear. Methods Human placenta and fetal cord blood samples of IUGR fetuses were collected, and the related indexes were detected. Pregnant Wistar rats were administered with 30 mg/kg·d (low dose) and 120 mg/kg·d (high dose) caffeine from gestational day (GD) 9 to 20 to construct the rat IUGR model. Pregnant mice were administered with caffeine (120 mg/kg·d) separately or combined with sodium ferulate (50 mg/kg·d) from gestational day GD 9 to 18 to confirm the intervention target on fetal weight loss caused by prenatal caffeine exposure (PCE). The fetal serum/placental corticosterone level, placental P-gp expression, and related indicator changes were analyzed. In vitro, primary human trophoblasts and BeWo cells were used to confirm the effect of caffeine on P-gp and its mechanism. Results The placental P-gp expression was significantly reduced, but the umbilical cord blood cortisol level was increased in clinical samples of the IUGR neonates, which were positively and negatively correlated with the neonatal birth weight, respectively. Meanwhile, in the PCE-induced IUGR rat model, the placental P-gp expression of IUGR rats was decreased while the corticosterone levels of the placentas/fetal blood were increased, which were positively and negatively correlated with the decreased placental/fetal weights, respectively. Combined with the PCE-induced IUGR rat model, in vitro caffeine-treated placental trophoblasts, we confirmed that caffeine decreased the histone acetylation and expression of P-gp via RYR/JNK/YB-1/P300 pathway, which inhibited placental and fetal development. We further demonstrated that P-gp inducer sodium ferulate could reverse the inhibitory effect of caffeine on the fetal body/placental weight. Finally, clinical specimens and other animal models of IUGR also confirmed that the JNK/YB-1 pathway is a co-regulatory mechanism of P-gp expression inhibition, among which the expression of YB-1 is the most stable. Therefore, we proposed that YB-1 could be used as the potential early warning target for the opening of the placental glucocorticoid barrier, the occurrence of IUGR, and the susceptibility of a variety of diseases. Conclusions This study, for the first time, clarified the critical role and epigenetic regulation mechanism of P-gp in mediating the opening mechanism of the placental glucocorticoid barrier, providing a novel idea for exploring the early warning, prevention, and treatment strategies of IUGR.

Pattern-based labeling methods have achieved promising results in alleviating the inevitable labeling noises of distantly supervised neural relation extraction. However, these methods require significant expert labor to write relation-specific patterns, which makes them too sophisticated to generalize quickly. To ease the labor-intensive workload of pattern writing and enable the quick generalization to new relation types, we propose a neural pattern diagnosis framework, DIAG-NRE, that can automatically summarize and refine highquality relational patterns from noise data with human experts in the loop. To demonstrate the effectiveness of DIAG-NRE, we apply it to two real-world datasets and present both significant and interpretable improvements over state-of-the-art methods. Source codes and data can be found at https://github. com/thunlp/DIAG-NRE.

No abstract

No abstract