Linguistic knowledge is of great benefit to scene text recognition. However, how to effectively model linguistic rules in end-to-end deep networks remains a research challenge. In this paper, we argue that the limited capacity of language models comes from: 1) implicitly language modeling; 2) unidirectional feature representation; and 3) language model with noise input. Correspondingly, we propose an autonomous, bidirectional and iterative ABINet for scene text recognition. Firstly, the autonomous suggests to block gradient flow between vision and language models to enforce explicitly language modeling. Secondly, a novel bidirectional cloze network (BCN) as the language model is proposed based on bidirectional feature representation. Thirdly, we propose an execution manner of iterative correction for language model which can effectively alleviate the impact of noise input. Additionally, based on the ensemble of iterative predictions, we propose a self-training method which can learn from unlabeled images effectively. Extensive experiments indicate that ABINet has superiority on lowquality images and achieves state-of-the-art results on several mainstream benchmarks. Besides, the ABINet trained with ensemble self-training shows promising improvement in realizing human-level recognition. Code is available at https://github.com/FangShancheng/ABINet.

Triclosan (TCS) is a high-volume chemical used as an antimicrobial ingredient in more than 2000 consumer products, such as toothpaste, cosmetics, kitchenware, and toys. We report that brief exposure to TCS, at relatively low doses, causes low-grade colonic inflammation, increases colitis, and exacerbates colitis-associated colon cancer in mice. Exposure to TCS alters gut microbiota in mice, and its proinflammatory effect is attenuated in germ-free mice. In addition, TCS treatment increases activation of Toll-like receptor 4 (TLR4) signaling in vivo and fails to promote colitis in mice. Together, our results demonstrate that this widely used antimicrobial ingredient could have adverse effects on colonic inflammation and associated colon tumorigenesis through modulation of the gut microbiota and TLR4 signaling. Together, these results highlight the need to reassess the effects of TCS on human health and potentially update policies regulating the use of this widely used antimicrobial.

Colon cancer is the third most common cancer and the second leading cause of cancer-related death in the United States, emphasizing the need for the discovery of new cellular targets. Using a metabolomics approach, we report here that epoxygenated fatty acids (EpFA), which are eicosanoid metabolites produced by cytochrome P450 (CYP) monooxygenases, were increased in both the plasma and colon of azoxymethane (AOM)/dextran sodium sulfate (DSS)-induced colon cancer mice. CYP monooxygenases were overexpressed in colon tumor tissues and colon cancer cells. Pharmacologic inhibition or genetic ablation of CYP monooxygenases suppressed AOM/DSS-induced colon tumorigenesis in vivo. In addition, treatment with 12,13-epoxyoctadecenoic acid (EpOME), which is a metabolite of CYP monooxygenase produced from linoleic acid, increased cytokine production and JNK phosphorylation in vitro and exacerbated AOM/DSS-induced colon tumorigenesis in vivo.Together, these results demonstrate that the previously unappreciated CYP monooxygenase pathway is upregulated in colon cancer, contributes to its pathogenesis, and could be therapeutically explored for preventing or treating colon cancer.Significance: This study finds that the previously unappreciated CYP monooxygenase eicosanoid pathway is deregulated in colon cancer and contributes to colon tumorigenesis. Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): Figure 4. EpOME exaggerates AOM/DSS-induced colon tumorigenesis in vivo. A, Scheme of animal experiment to test the effect of 12,13-EpOME (dose, 2 mg/kg/day; administered via mini-pump) on colon tumorigenesis. B, Quantification of colon tumorigenesis in mice (n ¼ 8-9 per group). C, Expression of proinflammatory and protumorigenic genes in colon (n ¼ 6-8 per group). D, Quantification of CD45 þ and CD45 þ F4/80 þ immune cells in colon (n ¼ 7-8 per group). E, Hematoxylin and eosin (H&E) histology and IHC staining of PCNA and b-catenin in colon (n ¼ 6-7 per group; scale bar, 50 mm). The results are expressed as mean AE SEM. The statistical significance of two groups was determined using Student t test or Wilcoxon-Mann test.