Convolutional neural networks (CNNs) with deep architectures have substantially advanced the state-of-the-art in computer vision tasks. However, deep networks are typically resource-intensive and thus difficult to be deployed on mobile devices. Recently, CNNs with binary weights have shown compelling efficiency to the community, whereas the accuracy of such models is usually unsatisfactory in practice. In this paper, we introduce network sketching as a novel technique of pursuing binary-weight CNNs, targeting at more faithful inference and better trade-off for practical applications. Our basic idea is to exploit binary structure directly in pre-trained filter banks and produce binaryweight models via tensor expansion. The whole process can be treated as a coarse-to-fine model approximation, akin to the pencil drawing steps of outlining and shading. To further speedup the generated models, namely the sketches, we also propose an associative implementation of binary tensor convolutions. Experimental results demonstrate that a proper sketch of AlexNet (or ResNet) outperforms the existing binary-weight models by large margins on the ImageNet large scale classification task, while the committed memory for network parameters only exceeds a little.
In this paper, we propose an alternative method to estimate room layouts of cluttered indoor scenes. This method enjoys the benefits of two novel techniques. The first one is semantic transfer (ST), which is: (1) a formulation to integrate the relationship between scene clutter and room layout into convolutional neural networks; (2) an architecture that can be end-to-end trained; (3) a practical strategy to initialize weights for very deep networks under unbalanced training data distribution. ST allows us to extract highly robust features under various circumstances, and in order to address the computation redundance hidden in these features we develop a principled and efficient inference scheme named physics inspired optimization (PIO). PIO's basic idea is to formulate some phenomena observed in ST features into mechanics concepts. Evaluations on public datasets LSUN and Hedau show that the proposed method is more accurate than state-of-the-art methods.
Summary
Astroviruses are a non‐enveloped virus with large host range breadth. AstV‐associated gastroenteritis in human and animal, nephritis in chicken, gout in gosling and hepatitis in duckling pose great threats to public health and poultry industry. Since early 2020, continuous emergence of fatal goose astrovirus (GAstV) infections characterized by articular and visceral gout was reported in China. Here, we described two outbreaks of emerging gout disease in two different goose farms of central China. Two virulent GAstV strains, designated as HNKF‐1/China/2020 and HNSQ‐6/China/2020, were isolated, and the fifth passage of the isolates could cause urate crystals accumulated in the allantoic fluid and even deposited around great vessels and embryo bodies. Meanwhile, the source of these GAstV outbreaks was tracked to goose hatcheries. The prevalence of GAstV in the goose embryos with hatch failure was confirmed, and embryo‐origin HNXX‐6/China/2020 was further isolated. The complete genome of these three newly isolates was then sequenced and analysed. The results showed that Chinese GAstVs have formed two distinct groups, and the three GAstV isolates, as well as most of the Chinese GAstVs, belong to the G‐I group. There are several amino acid mutations in the three newly identified GAstVs, such as A520T, S535R, V555I and A782T in ORF1a and Q229P in ORF2, suggesting the field stains, HNKF‐1/China/2020 and HNSQ‐6/China/2020, might derive from the weak goose embryo via vertical transmission. Moreover, the phylogenetic analysis of the complete viral genome and individual viral proteins revealed that Chinese GAstV strains have been constantly evolving towards more complicated and various directions. Our study reported the recently emerging GAstV outbreaks in central China, and further analysed the genetic characteristics of three virulent G‐I GAstV isolates from commercial goose farms and goose hatchery, indicating the diverse transmission of the virus and providing a basis for developing effective preventive measures and control strategies.
Class-imbalance extent metrics measure how imbalanced the data are. In pattern classification, it is usually expected that the higher the imbalance extent, the worse the classification performance, and thus an appropriate imbalance extent metric should show a negative correlation with the classification performance. Existing metrics, such as the popular imbalance ratio (IR), only consider the e↵ect of the sample sizes of di↵erent classes. However, we note that the dimensionality of imbalanced data also a↵ects the classification performance. Datasets with the same IR can present distinct classification performances when their dimensionalities are di↵erent, making IR suboptimal to reflect the imbalance extent for classification. We also observe that the classification performance becomes better with more discriminative features. Inspired by these observations, we propose a new imbalance extent metric, the adjusted IR, by adding a penalty term of the number of discriminative features that is e↵ectively determined by the Pearson correlation test. The adjusted IR adaptively revises the IR when the number of discriminative features varies. The empirical studies demonstrate the e↵ectiveness of the adjusted IR, in terms of its better negative correlation with the classification performance.
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.