To assess the vulnerability of deep learning in physical world, recent works introduce adversarial patches and apply them on different tasks. In this paper, we propose another kind patch: Meaningful Adversarial Sticker, a physically feasible stealthy attack method by using real stickers existing our life. Unlike previous designing perturbations, manipulates sticker's pasting position rotation angle objects to perform attacks. Because are less affected printing loss color distortion, can keep...

Human brains are commonly modeled as networks of Regions Interest (ROIs) and their connections for the understanding brain functions mental disorders. Recently, Transformer-based models have been studied over different types data, including graphs, shown to bring performance gains widely. In this work, we study network analysis. Driven by unique properties model graphs with nodes fixed size order, which allows us (1) use connection profiles node features provide natural low-cost positional...

This article proposes a novel differential evolution algorithm for solving constrained multimodal multiobjective optimization problems (CMMOPs), which may have multiple feasible Pareto-optimal solutions with identical objective vectors. In CMMOPs, due to the coexistence of multimodality and constraints, it is difficult current algorithms perform well in both decision spaces. The proposed uses speciation mechanism induce niches preserving more adopts an improved environment selection...

One of the most important food crops is rice. For this reason, accurate identification rice pests a critical foundation for pest control. In study, we propose an algorithm automatic and classification based on fully convolutional networks (FCNs) select 10 experiments. First, introduce new encoder–decoder in FCN series sub-networks connected by jump paths that combine long jumps shortcut connections fine-grained insect boundary detection. Secondly, network also integrates conditional random...

Adversarial patch is an important form of real-world adversarial attack that brings serious risks to the robustness deep neural networks. Previous methods generate patches by either optimizing their perturbation values while fixing pasting position or manipulating patch's content. This reveals positions and perturbations are both attack. For that, in this article, we propose a novel method simultaneously optimize for patch, thus obtain high success rate black-box setting. Technically, regard...

Recent studies on face forgery detection have shown satisfactory performance for methods involved in training datasets, but are not ideal enough unknown domains. This motivates many works to improve the generalization, forgery-irrelevant information, such as image background and identity, still exists different domain features causes unexpected clustering, limiting generalization. In this paper, we propose a controllable guide-space (GS) method enhance discrimination of domains, so increase...

Functional magnetic resonance imaging (fMRI) is one of the most common modalities to investigate brain functions. Recent studies in neuroscience stress great potential functional networks constructed from fMRI data for clinical predictions. Traditional networks, however, are noisy and unaware downstream prediction tasks, while also incompatible with deep graph neural network (GNN) models. In order fully unleash power GNNs network-based analysis, we develop FBNETGEN, a task-aware...

Introduction Fake news spread in various areas has a major negative impact on social life. Meanwhile, fake with text and visual content is more compelling than text-only quickly spreads across media. Therefore, detecting pressing task for the current society. Methods Concern problem of extracting insufficient features, inability to merge multi-modality features effectively news. In this article, we propose method by fusing data. Firstly, use two-branch learn hidden layer information modality...

Plant sensors are commonly used in agricultural production, landscaping, and other fields to monitor plant growth environmental parameters. As an important basic parameter monitoring, leaf inclination angle (LIA) not only influences light absorption pesticide loss but also contributes genetic analysis phenotypic data collection. The measurements of LIA provide a basis for crop research as well management, such water loss, absorption, illumination radiation. On the one hand, existing...

Background Recent studies have shown that deep learning can help tumor staging automatically. However, automatic nasopharyngeal carcinoma (NPC) is difficult due to the lack of large and slice‐level annotated datasets. Purpose To develop a weakly‐supervised deep‐learning method predict NPC patients' T stage without additional annotations. Study Type Retrospective. Population/Subjects In all, 1138 cases with from 2010 2012 were enrolled, including training set ( n = 712) validation 426). Field...

As the advance of multimedia technology, sources information are not limited to one form but emerging multi-modal properties. Through data processing, it is an essential component how detect fake news avoid their spread. The can leverage contents fabricate evidences or mislead readers, which damages a lot for management in social networks. In this work, we explore task multimodal detection. major challenge detection stems from modality fusion by abundant information. Overcoming limitations...

Abstract Foggy weather can cause such problems as blurred image information and the loss of details, which may pose great challenges to road traffic target detection based on images videos. In this study, we propose a domain‐adaptive vehicle method implement domain adaptation for real foggy scene. We firstly constructed highway dataset with (HVFD), contains normal provides complete data support computer vision. Secondly, by improving CycleGAN designed an improved generative confrontation...

With the advancement of deep learning technology, importance utilizing for livestock management is becoming increasingly evident. goat face detection provides a foundation recognition and management. In this study, we proposed novel neural network specifically designed object detection, addressing challenges such as low image resolution, small targets, indistinct features. By incorporating contextual information feature-fusion complementation, our approach was compared with existing networks...

Deep learning is the core technology of artificial intelligence, which has higher accuracy than traditional algorithms. The characteristics high-risk and high-yield in stock market make investors hope to predictions on it through scientific methods, so as reduce investment risks. Long short-term memory (LSTM) model deep can effectively describe long data suitable for predicting financial time series. Therefore, this paper uses LSTM learn forecast valuation indicator, price-earnings ratio...

The accurate and reproducible delineation of tumors from uninvolved tissue is essential for radiation oncology. However, the tumor margin may be challenging to identify magnetic resonance (MR) images nasopharyngeal carcinomas (NPCs). Additionally, clinical diagnoses such as T-staging already provide some information on invasion. To use this improve performance segmentation, we propose a novel deep learning neural network architecture that can incorporate both image information. Based U-Net,...

Precise weed recognition is an important step towards achieving intelligent agriculture. In this paper, a novel model, Cotton Weed-YOLO, proposed to improve the accuracy and efficiency of detection. CW-YOLO based on YOLOv8 introduces dual-branch structure combining Vision Transformer Convolutional Neural Network address problems small receptive field CNN high computational complexity transformer. The Receptive Field Enhancement (RFE) module enable feature pyramid network adapt information...