With the success of deep learning algorithms in various domains, studying adversarial attacks to secure deep models in real world applications has become an important research topic. Backdoor attacks are a form of adversarial attacks on deep networks where the attacker provides poisoned data to the victim to train the model with, and then activates the attack by showing a specific small trigger pattern at the test time. Most state-of-the-art backdoor attacks either provide mislabeled poisoning data that is possible to identify by visual inspection, reveal the trigger in the poisoned data, or use noise to hide the trigger. We propose a novel form of backdoor attack where poisoned data look natural with correct labels and also more importantly, the attacker hides the trigger in the poisoned data and keeps the trigger secret until the test time. We perform an extensive study on various image classification settings and show that our attack can fool the model by pasting the trigger at random locations on unseen images although the model performs well on clean data. We also show that our proposed attack cannot be easily defended using a state-of-the-art defense algorithm for backdoor attacks.
Landslide susceptibility zonation (LSZ) is necessary for disaster management and planning development activities in mountainous regions. A number of methods, viz. landslide distribution, qualitative, statistical and distribution-free analyses have been used for the LSZ studies and they are again briefly reviewed here. In this work, two methods, the Information Value (InfoVal) and the Landslide Nominal Susceptibility Factor (LNSF) methods that are based on bivariate statistical analysis have been applied for LSZ mapping in a part of the Himalayas. Relevant thematic maps representing various factors (e.g., slope, aspect, relative relief, lithology, buffer zones along thrusts, faults and lineaments, drainage density and landcover) that are related to landslide activity, have been generated using remote sensing and GIS techniques. The LSZ derived from the LNSF method, has been compared with that produced from the InfoVal method and the result shows a more realistic LSZ map from the LNSF method which appears to conform to the heterogeneity of the terrain.
The unprecedented success of deep neural networks in many applications has made these networks a prime target for adversarial exploitation. In this paper, we introduce a benchmark technique for detecting backdoor attacks (aka Trojan attacks) on deep convolutional neural networks (CNNs). We introduce the concept of Universal Litmus Patterns (ULPs), which enable one to reveal backdoor attacks by feeding these universal patterns to the network and analyzing the output (i.e., classifying the network as 'clean' or 'corrupted'). This detection is fast because it requires only a few forward passes through a CNN. We demonstrate the effectiveness of ULPs for detecting backdoor attacks on thousands of networks with different architectures trained on four benchmark datasets, namely the German Traffic Sign Recognition Benchmark (GTSRB), MNIST, CIFAR10, and Tiny-ImageNet. The codes and train/test models for this paper can be found here: https://umbcvision. github.io/Universal-Litmus-Patterns/.
The aim of the present study is to evaluate plant growth promoting and biocontrol efficacy of a Serratia marcescens strain ETR17 isolated from tea rhizosphere for the effective management of root rot disease in tea. Isolated bacterial culture ETR17 showed significant level of in vitro antagonism against nine different foliar and root pathogens of tea. The phenotypic and molecular characterization of ETR17 revealed the identity of the bacterium as Serratia marcescens. The bacterium was found to produce several hydrolytic enzymes like chitinase, protease, lipase, cellulase and plant growth promoting metabolites like IAA and siderophore. Scanning electron microscopic studies on the interaction zone between pathogen and antagonistic bacterial isolate revealed severe deformities in the fungal mycelia. Spectral analyses (LC-ESI-MS, UV-VIS spectrophotometry and HPLC) and TLC indicated the presence of the antibiotics pyrrolnitrin and prodigiosin in the extracellular bacterial culture extracts. Biofilm formation by ETR17 on polystyrene surface was also observed. In vivo application of talc-based formulations prepared with the isolate ETR17 in tea plantlets under green house conditions revealed effective reduction of root-rot disease as well as plant growth promotion to a considerable extent. Viability studies with the ETR17 talc formulation showed the survivability of the isolate up to six months at room temperature. The sustenance of ETR17 (concentration of 8-9x108 cfu g-1) in the soil after the application of talc formulation was recorded by ELISA. Safety studies revealed that ETR17 did not produce hemolysin as observed in pathogenic Serratia strains. The biocontrol strain reported in this study can be used for field application in order to minimize the use of chemical fungicides for disease control in tea gardens.
With the success of deep learning algorithms in various domains, studying adversarial attacks to secure deep models in real world applications has become an important research topic. Backdoor attacks are a form of adversarial attacks on deep networks where the attacker provides poisoned data to the victim to train the model with, and then activates the attack by showing a specific small trigger pattern at the test time. Most state-of-the-art backdoor attacks either provide mislabeled poisoning data that is possible to identify by visual inspection, reveal the trigger in the poisoned data, or use noise to hide the trigger. We propose a novel form of backdoor attack where poisoned data look natural with correct labels and also more importantly, the attacker hides the trigger in the poisoned data and keeps the trigger secret until the test time.We perform an extensive study on various image classification settings and show that our attack can fool the model by pasting the trigger at random locations on unseen images although the model performs well on clean data. We also show that our proposed attack cannot be easily defended using a state-of-the-art defense algorithm for backdoor attacks.
The benefits of utilizing spatial context in fast object detection algorithms have been studied extensively. Detectors increase inference speed by doing a single forward pass per image which means they implicitly use contextual reasoning for their predictions. However, one can show that an adversary can design adversarial patches which do not overlap with any objects of interest in the scene and exploit contextual reasoning to fool standard detectors. In this paper, we examine this problem and design category specific adversarial patches which make a widely used object detector like YOLO blind to an attacker chosen object category. We also show that limiting the use of spatial context during object detector training improves robustness to such adversaries. We believe the existence of context based adversarial attacks is concerning since the adversarial patch can affect predictions without being in vicinity of any objects of interest. Hence, defending against such attacks becomes challenging and we urge the research community to give attention to this vulnerability.
In an approach toward the development of ecofriendly antifungal compounds for controlling major foliar fungal diseases of tea, ethanol and aqueous extracts of 30 plants belonging to 20 different families collected from sub-Himalayan West Bengal (India) were tested against the pathogens Pestalotiopsis theae (Saw.) Stey., Colletotrichum camelliae Mess., Curvularia eragrostidis (P. Hennings) Meyer, and Botryodiplodia theobromae Patouiilard. Spore germination technique was followed for evaluation of antifungal properties. Results showed that ethanol and aqueous extracts of Allium sativum L., Datura metel L., Dryopteris filix-mas (L.) Schott, Zingiber officinale Rosc., Smilax zeylanica L., Azadirachta indica, A. Joss. and Curcuma longa L. recorded 100% inhibition of spore germination. The antifungal component from these plants may be used in developing novel fungicides for tea gardens.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
334 Leonard St
Brooklyn, NY 11211
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.