Search Results (8)

The high demands on datasets and computing resources in deep learning make the models vulnerable to a range of security threats such as backdoor learning. The study of backdoor learning also helps to improve the understanding of model security. In order to ensure the attack effect, the triggers and targets in the existing backdoor learning methods are usually fixed and single, so a single defense will lead to the failure of the attack. This paper proposes an invisible backdoor learning scheme in the transform domain with flexible triggers and targets. By adding different offsets of different frequencies in the transform domain, multiple triggers and multiple targets are controlled. The generated poisoning images are added to the training dataset and the model is fine-tuned. Under the conception, two modes of backdoor learning enable flexible triggers and targets. One mode is multi-triggers and multi-targets (MTMT), and it can implement multiple triggers corresponding to different activation targets. The other mode is multi-triggers and one-target (MTOT), and it can realize multiple trigger sets to activate the target together. The experimental results show that the attack success rate reaches 95% and the accuracy of the model decreases within 3% under the premise that the trigger is not visible. This scheme can resist the common defense methods and has a good sample of the visual quality. Full article

Backdoor attacks aim to implant hidden backdoors into Deep Neural Networks (DNNs) so that the victim models perform well on clean images, whereas their predictions would be maliciously changed on poisoned images. However, most existing backdoor attacks lack the invisibility and robustness required for real-world applications, especially when it comes to resisting image compression techniques, such as JPEG and WEBP. To address these issues, in this paper, we propose a Backdoor Attack Method based on Trigger Generation (BATG). Specifically, a deep convolutional generative network is utilized as the trigger generation model to generate effective trigger images and an Invertible Neural Network (INN) is utilized as the trigger injection model to embed the generated trigger images into clean images to create poisoned images. Furthermore, a noise layer is used to simulate image compression attacks for adversarial training, enhancing the robustness against real-world image compression. Comprehensive experiments on benchmark datasets demonstrate the effectiveness, invisibility, and robustness of the proposed BATG. Full article

Deep Generative Models (DGMs), as a state-of-the-art technology in the field of artificial intelligence, find extensive applications across various domains. However, their security concerns have increasingly gained prominence, particularly with regard to invisible backdoor attacks. Currently, most backdoor attack methods rely on visible backdoor triggers that are easily detectable and defendable against. Although some studies have explored invisible backdoor attacks, they often require parameter modifications and additions to the model generator, resulting in practical inconveniences. In this study, we aim to overcome these limitations by proposing a novel method for invisible backdoor attacks. We employ an encoder–decoder network to ‘poison’ the data during the preparation stage without modifying the model itself. Through meticulous design, the trigger remains visually undetectable, substantially enhancing attacker stealthiness and success rates. Consequently, this attack method poses a serious threat to the security of DGMs while presenting new challenges for security mechanisms. Therefore, we urge researchers to intensify their investigations into DGM security issues and collaboratively promote the healthy development of DGM security. Full article

In recent years, plastic pollution has become a growing environmental concern: more than 350 million tons of plastic material are produced annually. Although many efforts have been made to recycle waste, a significant proportion of these plastics contaminate and accumulate in the environment. A central point in plastic pollution is demonstrated by the evidence that plastic objects gradually and continuously split up into smaller pieces, thus producing subtle and invisible pollution caused by microplastics (MP) and nanoplastics (NP). The small dimensions of these particles allow for the diffusion of these contaminants in farmlands, forest, freshwater, and oceans worldwide, posing serious menaces to human, animal, and plant health. The uptake of MPs and NPs into plant cells seriously affects plant growth, development, and photosynthesis, finally limiting crop yields and endangering natural environmental biodiversity. Furthermore, nano- and microplastics—once adsorbed by plants—can easily enter the food chain, being highly toxic to animals and humans. This review addresses the impacts of MP and NP particles on plants in the terrestrial environment. In particular, we provide an overview here of the detrimental effects of photosynthetic injuries, oxidative stress, ROS production, and protein damage triggered by MN and NP in higher plants and, more specifically, in crops. The possible damage at the physiological and environmental levels is discussed. Full article

Based on helioseismological measurements (1996–2017), the entire Sun shrinks during solar maximum and regrows during the next solar minimum by about a few km (~10⁻⁵ effect). Here, we observe, for the first time, that the solar radius variation resembles a 225-day relationship that coincides with Venus’ orbital period. We show that a remote link between planet Venus and Sun’s size must be at work. However, within known realms of physics, this is unexpected. Therefore, we can only speculate about its cause. Notably, the driving idea behind this investigation was some generic as-yet-invisible matter from the dark Universe. In fact, the 11-year solar cycle shows planetary relationships for a number of other observables as well. It has been proposed that the cause must be due to some generic streaming invisible massive matter (IMM). As when a low-speed stream is aligned toward the Sun with an intervening planet, the IMM influx increases temporally due to planetary gravitational focusing, assisted eventually with the free fall of incident slow IMM. A case-specific simulation for Venus’ impact supports the tentative scenario based on this investigation’s driving idea. Importantly, Saturn, combined with the innermost planets Mercury or Venus, unambiguously confirms an underlying planetary correlation with the Sun’s size. The impact of the suspected IMM accumulates with time, slowly triggering the underlying process(es); the associated energy change is massive even though it extends from months to several years. This study shows that the Sun’s size response is as short as half the orbital period of Mercury (44 days) or Venus (112 days). Then, the solar system is the target and the antenna of still unidentified external impact, assuming tentatively from the dark sector. If the generic IMM also has some preferential incidence direction, future long-lasting observations of the Sun’s shape might provide an asymmetry that could be utilized to identify the not isotropic influx of the assumed IMM. Full article

The construction industry is one of the most hazardous industries in many countries. Many studies have asserted that industrial accidents could be prevented by eliminating their root causes. However, given that accident occurrence processes are considerably complex and often invisible, it is difficult to identify and eliminate the root causes. Based on this recognition, this paper aims to analyze the causality of construction accidents on the basis of direct causes that are classified into unsafe actions (UA) and unsafe conditions (UC). A logistic regression is applied to examine associations between UAs and UCs and their significances in triggering construction accidents. Then, a Delphi method is applied to determine the relationships between direct and root causes of construction accidents. This study contributes to an improved understanding of the complex causal process of construction accidents, which is a necessary stepping-stone to prevent construction accidents. Meanwhile, only one-to-one combinations of UCs and UAs are considered in this paper. Thus, follow-up studies to examine the impact of one-to-many or many-to-many combinations are needed. Full article

Gravitational waves astronomy allows us to study objects and events invisible in electromagnetic waves. It is crucial to validate the theories and models of the most mysterious and extreme matter in the Universe: the neutron stars. In addition to inspirals and mergers of neutrons stars, there are currently a few proposed mechanisms that can trigger radiation of long-lasting gravitational radiation from neutron stars, such as e.g., elastically and/or magnetically driven deformations: mountains on the stellar surface supported by the elastic strain or magnetic field, free precession, or unstable oscillation modes (e.g., the r-modes). The astrophysical motivation for continuous gravitational waves searches, current LIGO and Virgo strategies of data analysis and prospects are reviewed in this work. Full article

Radon is a noble gas originating from the radioactive decay chain of uranium or thorium. Most radon emanates naturally from the soil and from some building materials, so it can be found in many places around the world, in particular in regions with soils containing granite or slate. It is almost impossible for a person to detect radon gas without proper tools, since it is invisible, odorless, tasteless and colorless. The problem is that a correlation has been established between the presence of high radon gas concentrations and the incidence of lung cancer. In fact, the World Health Organization (WHO) has stated that the exposure to radon is the second most common cause of lung cancer after smoking, and it is the primary cause of lung cancer among people who have never smoked. Although there are commercial radon detectors, most of them are either expensive or provide very limited monitoring capabilities. To tackle such an issue, this article presents a cost-effective IoT radon gas remote monitoring system able to obtain accurate concentration measurements. It can also trigger events to prevent dangerous situations and to warn users about them. Moreover, the proposed solution can activate mitigation devices (e.g., forced ventilation) to decrease radon gas concentration. In order to show its performance, the system was evaluated in three different scenarios corresponding to representative buildings in Galicia (Spain), a region where high radon gas concentrations are common due to the composition of the soil. In addition, the influence of using external hardware (i.e., WiFi transceivers and an embedded System-on-Chip (SoC)) next to the radon gas sensor is studied, concluding that, in the tested scenarios, they do not interfere with the measurements. Full article