Training neural networks with good generalization requires large computational costs in many deep learning methods due to large-scale datasets and over-parameterized models. Despite the emergence of a number of coreset selection methods to reduce the computational costs, the problem of coreset distribution bias, i.e., the skewed distribution between the coreset and the entire dataset, has not been well studied. In this paper, we find that the closer the feature distribution of the coreset is to that of the entire dataset, the better the generalization performance of the coreset, particularly under extreme pruning. This motivates us to propose a simple yet effective method for coreset selection to alleviate the distribution bias between the coreset and the entire dataset, called feature distribution matching (FDMat). Unlike gradient-based methods, which selects samples with larger gradient values or approximates gradient values of the entire dataset, FDMat aims to select coreset that is closest to feature distribution of the entire dataset. Specifically, FDMat transfers coreset selection as an optimal transport problem from the coreset to the entire dataset in feature embedding spaces. Moreover, our method shows strong robustness due to the removal of samples far from the distribution, especially for the entire dataset containing noisy and class-imbalanced samples. Extensive experiments on multiple benchmarks show that FDMat can improve the performance of coreset selection than existing coreset methods. The code is available at https://github.com/successhaha/FDMat.

When a disaster strikes, accurate disaster information and effective response are critical for saving lives and properties. High-resolution satellite (HRS) imagery provides valuable geographical information that can assist experts in analyzing damage levels in different areas and enacting appropriate relief plans. However, analyzing large HRS images is both time-consuming and inefficient, requiring efficient automated methods to replace expert analysis. Fortunately, deep learning methods have achieved impressive performance on HRS image processing tasks, considerably increasing automation levels. Despite this progress, most HRS-based damage assessment methods only consider a single time series of post-disaster images or simply integrate pre- and post-disaster images, lacking the integration of effective information between pre- and post-disaster images. To alleviate this problem, we propose a two-stage multi-scale fusion network that fully exploits the information contained in pre- and post-disaster images. Specifically, we employ a hierarchical Transformer to accurately locate buildings by pre-disaster images in the first stage, and then propose the guided fusion and cross-scale guided fusion modules in the second stage to efficiently utilize both pre- and post-disaster images. Our method outperforms state-of-the-art methods in building segmentation and building damage assessment on the xBD dataset, and exhibits improved generalization across diverse geographic regions and disaster types.

This paper introduces the research status of the business model at home and abroad, including the concept of business model, business model classification, the relationship between the business model classification and enterprise performance, and puts forward the shortcomings of study on the business model classification. Based on previous studies, this paper puts forward a new business model classification according to Chinese IT industry characters. Then enterprise performance is divided into four aspects: profitability, growth, operating efficiency and market value. This paper introduces a single-factor analysis of variance, and the process of studying the effects of different business models on the various aspects of enterprise performance in order to find a better business model, and the method can provide recommendations for the enterprise business model transformation.

This paper studies a predicting model of the reverse logistics and its application in beer bottle recycling process. The model is composed of long-term growth trend and seasonal variation where long-term growth trend is described by gray model and seasonal variation is described by seasonal index. The reverse logistics value chain is analyzed by RMM. Then this paper studies two recovery paths of the beer bottle and all links of paths which reveals the application scenarios of the predicting model of the reverse logistics. Some suggestions are put forward in the end.

Image watermarking has become a popular technique for authentication and copyright protection. However, many proposed image watermarking techniques are sensitive to affine transforms, such as rotation, scaling and translation. Here, a local affine transform invariant watermarking is designed and tested against attacks performed by Stirmark using the Krawtchouk moment invariants. Watermark is inserted into the perceptually significant Krawtchouk moment invariants of the original image, and watermarking based on Krawtchouk moment invariants are local, that is, the embedded watermark affects only a selected portion of the original image, the position of which can be decided by the user. This, in effect, permits the watermark to be embedded at the portion of the image which is most significant information-wise. This also means that the watermark is especially robust to cropping. Independent component analysis (ICA) is utilised by detector to extract the perfect watermark blindly. The computational aspects of the proposed watermarking are also discussed in detail. Experimental results have demonstrated that the proposed watermarking technique has a good robustness against other attacks performed by Stirmark including affine transform, cropping, filtering, image compression and random geometric distortions. It is indicated that the proposed watermarking has superior advantages over the existing ones in many aspects.