Forest fire is a ubiquitous disaster which has long-term impact on the local climate as well ecological balance and products based remote sensing satellite data have developed rapidly. However, early forest smoke in images small area easily confused by clouds fog, makes it difficult to be identified. Too many redundant frequency bands index for will an interference wildfire detection, resulting decline detection accuracy efficiency smoke. To solve these problems, this study analyzed...

Most of the image encryption schemes based on chaos have so far employed symmetric key cryptography, which leads to a situation where cannot be transmitted in public channels, thus limiting their extended application. Based elliptic curve cryptography (ECC), we proposed method hash value derived from plain was encrypted by ECC. Furthermore, during permutation, novel algorithm different-sized block proposed. The firstly divided into five planes according amount information contained different...

Federated learning (FL) has achieved state-of-the-art performance in distributed tasks with privacy requirements. However, it been discovered that FL is vulnerable to adversarial attacks. The typical gradient inversion attacks primarily focus on attempting obtain the client's private input a white-box manner, where adversary assumed be either client or server. if both clients and server are honest fully trusted, secure? In this paper, we propose novel method called External Gradient...

Forest fires are one of the most devastating natural disasters, and technologies based on remote sensing satellite data for fire prevention control have developed rapidly in recent years. Early forest smoke images, other hand, is thin tiny area, making it difficult to detect. Satellites with high spatial resolution sensors can collect high-resolution photographs smoke, however impact satellite’s repeat access time same area means that cannot be detected time. Because their low resolution,...

Knowing model parameters has been regarded as a vital factor for recovering sensitive information from the gradients in federated learning. But is it safe to use learning when are unavailable adversaries, i.e., external adversaries' In this paper, we answer question by proposing novel gradient inversion attack. Speciffically, observe widely ignored fact that participants' data usually transmitted via intermediary node. Based on fact, show an adversary able recover private input gradients,...

The redactable blockchain has emerged as a promising technique in mobile crowdsensing, allowing users to break immutability controlled manner selectively. Unfortunately, current fine-grained blockchains suffer two significant limitations terms of security and functionality, which severely impede their application crowdsensing. For security, the transparency allows anyone access both data policy, consequently results breach user privacy. Regarding solutions cannot support error tolerance...

In NLP(Natural Language Processing), the direct object of computer processing is not actual natural language, but its computational model. Therefore, to truly understand problems NLP and find out solutions, we need discuss these from perspective language modeling. this paper, characteristics are analyzed, model, word segmentation, part-of-speech tagging named entity recognition in studied by using method based on RNN(Recursive Neural Network). A dependency parsing model LSTM proposed. The...

Abstract The Visual Question Answering (VQA) task is to infer the correct answer a free-form question based on given image. This challenging because it requires model handling both visual and textual information. Most successful attempts VQA have been achieved by using attention mechanism which can capture inter-modal intra-modal dependencies. In this paper, we raise new attention-based solve VQA. We use information guide concentrate special regions attribute hierarchically reason answer....

Ubiquitous intelligence empowered internet of vehicles (UIIoV) is an emerging paradigm where network entities such as mobile vehicles, edge/cloud servers, and intermediate nodes interact to achieve effective data sensing make intelligent decisions based on artificial techniques. Unlike the conventional introduced improve ability processing analysis, further enhance accuracy decision-making. In this article, we study architecture UIIoV, security, privacy, reliability challenges it confronts....

Recent studies have shed light on security vulnerabilities in Encoder-as-a-Service (EaaS) systems that enable the theft of valuable encoder attributes such as functionality. However, many these attacks often either simply used data augmentation method, or solely explored idea contrastive learning to improve performance, lacking analysis and a combination both two aspects. Furthermore, they also ignored potential harnessing inner characteristics encoder, specifically its robustness. Thus, we...

With the internet of things, autonomous driving, and smart security rapid development, data growth puts higher more urgent requirements on processor computing performance. A single chip is hard to meet real-time in many scenarios, so it necessary expand multiple chips. This paper proposes a module-level pipeline FPGA cluster structure based inter-board heterogeneous. The method divides enormous task into modules running each module different as between boards. In addition, considering impact...