Convolutional neural networks (ConvNets) have achieved excellent recognition performance in various visual recognition tasks. A large labeled training set is one of the most important factors for its success. However, it is difficult to collect sufficient training images with precise labels in some domains, such as apparent age estimation, head pose estimation, multilabel classification, and semantic segmentation. Fortunately, there is ambiguous information among labels, which makes these tasks different from traditional classification. Based on this observation, we convert the label of each image into a discrete label distribution, and learn the label distribution by minimizing a Kullback-Leibler divergence between the predicted and ground-truth label distributions using deep ConvNets. The proposed deep label distribution learning (DLDL) method effectively utilizes the label ambiguity in both feature learning and classifier learning, which help prevent the network from overfitting even when the training set is small. Experimental results show that the proposed approach produces significantly better results than the state-of-the-art methods for age estimation and head pose estimation. At the same time, it also improves recognition performance for multi-label classification and semantic segmentation tasks.
This paper aims at accelerating and compressing deep neural networks to deploy CNN models into small devices like mobile phones or embedded gadgets. We focus on filter level pruning, i.e., the whole filter will be discarded if it is less important. An effective and unified framework, ThiNet (stands for "Thin Net"), is proposed in this paper. We formally establish filter pruning as an optimization problem, and reveal that we need to prune filters based on statistics computed from its next layer, not the current layer, which differentiates ThiNet from existing methods. We also propose "gcos" (Group COnvolution with Shuffling), a more accurate group convolution scheme, to further reduce the pruned model size. Experimental results demonstrate the effectiveness of our method, which has advanced the state-of-the-art. Moreover, we show that the original VGG-16 model can be compressed into a very small model (ThiNet-Tiny) with only 2.66MB model size, but still preserve AlexNet level accuracy. This small model is evaluated on several benchmarks with different vision tasks (e.g., classification, detection, segmentation), and shows excellent generalization ability.
Background-Rad (Ras associated with diabetes) GTPase is the prototypic member of a subfamily of Ras-related small G proteins. The aim of the present study was to define whether Rad plays an important role in mediating cardiac hypertrophy. Methods and Results-We document for the first time that levels of Rad mRNA and protein were decreased significantly in human failing hearts (nϭ10) compared with normal hearts (nϭ3; PϽ0.01). Similarly, Rad expression was decreased significantly in cardiac hypertrophy induced by pressure overload and in cultured cardiomyocytes with hypertrophy induced by 10 mol/L phenylephrine. Gain and loss of Rad function in cardiomyocytes significantly inhibited and increased phenylephrine-induced hypertrophy, respectively. In addition, activation of calcium-calmodulin-dependent kinase II (CaMKII), a strong inducer of cardiac hypertrophy, was significantly inhibited by Rad overexpression. Conversely, downregulation of CaMKII␦ by RNA interference technology attenuated the phenylephrine-induced hypertrophic response in cardiomyocytes in which Rad was also knocked down. To further elucidate the potential role of Rad in vivo, we generated Rad-deficient mice and demonstrated that they were more susceptible to cardiac hypertrophy associated with increased CaMKII phosphorylation than wild-type littermate controls. Conclusions-The present data document for the first time that Rad is a novel mediator that inhibits cardiac hypertrophy through the CaMKII pathway. The present study will have significant implications for understanding the mechanisms of cardiac hypertrophy and setting the basis for the development of new strategies for treatment of cardiac hypertrophy.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.