Attention networks in multimodal learning provide an efficient way to utilize given visual information selectively. However, the computational cost to learn attention distributions for every pair of multimodal input channels is prohibitively expensive. To solve this problem, co-attention builds two separate attention distributions for each modality neglecting the interaction between multimodal inputs. In this paper, we propose bilinear attention networks (BAN) that find bilinear attention distributions to utilize given vision-language information seamlessly. BAN considers bilinear interactions among two groups of input channels, while low-rank bilinear pooling extracts the joint representations for each pair of channels. Furthermore, we propose a variant of multimodal residual networks to exploit eight-attention maps of the BAN efficiently. We quantitatively and qualitatively evaluate our model on visual question answering (VQA 2.0) and Flickr30k Entities datasets, showing that BAN significantly outperforms previous methods and achieves new state-of-the-arts on both datasets.

Epigalloccatechin-3-gallate (EGCG) is the main polyphenol component of green tea (leaves of Camellia sinensis). EGCG is known for its antioxidant, anti-inflammatory, antiviral, and anti-carcinogenic properties. Here, we identify EGCG as a new inhibitor of ocular angiogenesis and its vascular permeability. Matrix metalloproteinases (MMPs) and vascular endothelial growth factor (VEGF) play a key role in the processes of extracellular matrix (ECM) remodeling and microvascular permeability during angiogenesis. We investigated the inhibitory effects of EGCG on ocular neovascularization and vascular permeability using the retina oriented cells and animal models induced by VEGF and alkaline burn. EGCG treatment significantly decreased mRNA and protein expression levels of MMP-9 in the presence of 12-O-tetradecanoylphorbol-13-acetate (TPA) and tumor necrosis factor alpha (TNF-α) in human retinal pigment epithelial cells (HRPECs). EGCG also effectively protected ARPE-19 cells from cell death and attenuated mRNA expressions of key angiogenic factors (MMP-9, VEGF, VEGF Receptor-2) by inhibiting generation of reactive oxygen species (ROS). EGCG significantly inhibited proliferation, vascular permeability, and tube formation in VEGF-induced human retinal microvascular endothelial cells (HRMECs). Furthermore, EGCG significantly reduced vascular leakage and permeability by blood-retinal barrier breakdown in VEGF-induced animal models. OPEN ACCESSMolecules 2014, 19 12151In addition, EGCG effectively limited upregulation of MMP-9 and platelet endothelial cell adhesion molecule (PECAM/CD31) on corneal neovascularization (CNV) induced by alkaline burn. Our data suggest that MMP-9 and VEGF are key therapeutic targets of EGCG for treatment and prevention of ocular angiogenic diseases such as age-related macular degeneration, diabetic retinopathy, and corneal neovascularization.

No abstract

No abstract

No abstract