In order to solve the problems of long artificial time consumption, inability standardize degree damage, and difficulty maintaining data in traditional tunnel disease detection methods, this paper proposes use Residual Network (ResNet) models for water leakage crack detection. ResNet a residual learning framework ease training networks that are deeper than those previously used. Furthermore, explicitly reformulates layers as functions reference layer inputs, rather unreferenced functions....

Remote sensing image fusion (RSIF) can generate an integrated with high spatial and spectral resolution. The fused remote is conducive to applications including disaster monitoring, ecological environment investigation, dynamic monitoring. However, most existing deep learning based RSIF methods require ground truths (or reference images) train a model, the acquisition of difficult problem. To address this, we propose semisupervised method on multiscale conditional generative adversarial...

The widespread dissemination of high-fidelity fake faces created by face forgery techniques has caused serious trust concerns and ethical issues in modern society. Consequently, detection emerged as a prominent topic research to prevent technology abuse. Although, most existing detectors demonstrate success when evaluating high-quality under intra-dataset scenarios, they often overfit manipulation-specific artifacts lack robustness postprocessing operations. In this work, we design an...

Abstract Image hiding is a task that hides secret images into cover images. The purposes of image are to ensure the invisible human and can be recovered. current state‐of‐the‐art steganography methods run risk information leakage. A safe network (SIHNet) presented reduce leakage information. Based on some phenomena which use invertible neural network, reversible processing (SIP) module proposed make suitable for stego leak less Besides, lost (LIH) used hide images, thus method recover better...

For Fangshan granite in Beijing, the static compression and dynamic tests have been carried out separately under natural air drying water saturation. It was found that compressive strength of water-saturated is higher than air-dried granite, which contrary to result rock lower load. Furthermore, medium strain rate condition, when 85 s−1, could be increased by nearly 0.5 times compared with its state. The 1–2 strength, shows has stronger sensitivity granite. Meanwhile, impact loading, from...

Abstract Deep neural networks are widely adopted powerful tools for perceptual tasks. However, recent research indicated that they easily fooled by adversarial examples, which produced adding imperceptible perturbations to clean examples. Here the steganalysis rich model (SRM) is utilized generate noise feature maps, and combined with RGB images discover difference between examples In particular, a two‐stream pseudo‐siamese network fuses subtle in inconsistency features proposed. The...

Identifying the subcellular localization of a given protein is an essential part biological and medical research, since must be localized in correct organelle to ensure physiological function. Conventional experiments for have some limitations, such as high cost low efficiency, thus massive computational methods are proposed solve these problems. However, need improved further with class imbalance problem. We propose new model, generating minority samples (Gm-PLoc), predict multi-label...

With the popularity of posting memes on social platforms, severe negative impact hateful is growing. As existing detection models have lower accuracy than humans, still a challenge to statistical learning and artificial intelligence. This paper proposed multi-task method consisting primary multimodal task two unimodal auxiliary tasks address this issue. We introduced self-supervised generation strategy in generate labels automatically. Meanwhile, we used BERT RESNET as backbone for text...

Combining the traditional enterprise performance measurement method with nature and characteristics of integrated supply chain, this research determines scope contents chain system. In line individual theory, systems should reflect strategy chain. At same time, it be able to check efficiency strategy. According layer upon strategic objects five key aspects (finance, business processes, customer, environment, core ability) are found. These fields significant for formation improvement

With increasing prevalence of global virtual teams (GVTs), GVTs effectiveness draws much attention from both academia and industries. Communication is widely acknowledged as a significant element to improve the source many problems in GVTs, but there no systematic model about it. In this paper, conceptual proposed direct communication management CVTs. model, team characteristics are considered independent variables, including degree virtuality diversity. addition, factors influencing...

In recent years, tensor-based machine learning methods, in which the Support Tensor Machine (STM) is a typical technology, have gradually attracted attention of researchers. Compared with Vector (SVM), STM has superior generalization ability that can make full use structural information data. However, it still faces many challenges due to imperfection its theoretical basis and model. order study further development STM, this paper provides survey about potential existing problems STM.

Abstract Arbitrary style transfer aims to stylize the content image with image. The key problem of is how balance global structure and local patterns. A promising method solve this attentional method, where a learnable embedding features enables patterns be flexibly recombined image, so will well preserved in stylized However, current methods cannot preserve structure. To problem, novel network proposed, that relies on Optimal Transport (OT) for computing attention map. proposed OT‐based...

Adversarial examples have the property of transferring across models, which has created a great threat for deep learning models. To reveal shortcomings in existing method ensemble been introduced to generating transferable adversarial examples. However, most model attacks directly combine different models’ output but ignore large differences optimization direction them, severely limits transfer attack ability. In this work, we propose new kind called stochastic average attack. Unlike...

Due to the increasing sophistication of face forgery techniques, images generated are becoming more and realistic difficult for human eyes distinguish. These techniques can cause problems such as fraud social engineering attacks in facial recognition identity verification areas. Therefore, researchers have worked on detection studies made significant progress. Current algorithms achieve high accuracy within-dataset. However, it is satisfactory generalization performance cross-dataset...