No abstract

Background Ginseng ( Panax ginseng Meyer) is an invaluable medicinal plant containing various bioactive metabolites (e.g., ginsenosides). Owing to its long cultivation period, ginseng is vulnerable to various biotic constraints. Biological control using endophytes is an important alternative to chemical control. Methods In this study, endophytic Trichoderma citrinoviride PG87, isolated from mountain-cultivated ginseng, was evaluated for biocontrol activity against six major ginseng pathogens. T. citrinoviride exhibited antagonistic activity with mycoparasitism against all ginseng pathogens, with high endo-1,4-β-D-glucanase activity. Results T. citrinoviride inoculation significantly reduced the disease symptoms caused by Botrytis cinerea and Cylindrocarpon destructans and induced ginsenoside biosynthesis in ginseng plants. T. citrinoviride was formulated as dustable powder and granules. The formulated agents also exhibited significant biocontrol activity and induced ginsenosides production in the controlled environment and mountain area. Conclusion Our results revealed that T. citrinoviride has great potential as a biological control agent and elicitor of ginsenoside production.

In this study, we isolated a total of 238 culturable putative bacterial endophytes from four Pinus species ( Pinus densiflora , P . koraiensis , P . rigida , and P . thunbergii ) across 18 sampling sites in Korea. The samples were cultured in de Man Rogosa Sharpe and humic acid-vitamin agar media. These selective media were used to isolate lactic acid bacteria and Actinobacteria , respectively. Analysis using 16S ribosomal DNA sequencing grouped the isolated putative bacterial endophytes into 107 operational taxonomic units (OTUs) belonging to 48 genera. Gamma - proteobacteria were the most abundant bacteria in each sampling site and three tissues (needle, stem and root). The highest OTU richness and diversity indices were observed in the roots, followed by stem and needle tissues. Total metabolites extracted from three isolates (two isolates of Escherichia coli and Serratia marcescens ) showed significant nematicidal activity against the pine wood nematode ( Bursaphelenchus xylophilus ). Our findings demonstrated the potential use of bacterial endophytes from pine trees as alternative biocontrol agents against pine wood nematodes.

Adversarial examples are perturbed inputs that are designed to deceive machine-learning classifiers by adding adversarial perturbations to the original data. Although fast adversarial training have demonstrated both robustness and efficiency, the problem of "catastrophic overfitting" has been observed. It is a phenomenon that, during single-step adversarial training, the robust accuracy against projected gradient descent (PGD) suddenly decreases to 0% after few epochs, whereas the robustness against fast gradient sign method (FGSM) increases to 100%. In this paper, we address three main topics. (i) We demonstrate that catastrophic overfitting occurs in single-step adversarial training because it trains adversarial images with maximum perturbation only, not all adversarial examples in the adversarial direction, which leads to a distorted decision boundary and a highly curved loss surface. (ii) We experimentally prove this phenomenon by proposing a simple method using checkpoints. This method not only prevents catastrophic overfitting, but also overrides the belief that single-step adversarial training is hard to prevent multi-step attacks. (iii) We compare the performance of the proposed method to that obtained in recent works and demonstrate that it provides sufficient robustness to different attacks even after hundreds of training epochs in less time. All code for reproducing the experiments in this paper are at https://github.com/Harry24k/catastrophic-overfitting.

Although fast adversarial training has demonstrated both robustness and efficiency, the problem of "catastrophic overfitting" has been observed. This is a phenomenon in which, during single-step adversarial training, the robust accuracy against projected gradient descent (PGD) suddenly decreases to 0% after a few epochs, whereas the robust accuracy against fast gradient sign method (FGSM) increases to 100%. In this paper, we demonstrate that catastrophic overfitting is very closely related to the characteristic of single-step adversarial training which uses only adversarial examples with the maximum perturbation, and not all adversarial examples in the adversarial direction, which leads to decision boundary distortion and a highly curved loss surface. Based on this observation, we propose a simple method that not only prevents catastrophic overfitting, but also overrides the belief that it is difficult to prevent multi-step adversarial attacks with single-step adversarial training.