Search

Deep neural networks (DNNs) have achieved breakthrough progress in various fields, largely owing to the support of large-scale datasets with manually annotated labels. However, obtaining such datasets is costly and time-consuming, making high-quality annotation a challenging task. In this work, we propose an improved noisy sample selection method, termed "sample selection framework", based on a mixup loss and recalibration strategy (SMR). This framework enhances the robustness and generalization abilities of models. First, we introduce a robust mixup loss function to pre-train two models with identical structures separately. This approach avoids additional hyperparameter adjustments and reduces the need for prior knowledge of noise types. Additionally, we use a Gaussian Mixture Model (GMM) to divide the entire training set into labeled and unlabeled subsets, followed by robust training using semi-supervised learning (SSL) techniques. Furthermore, we propose a recalibration strategy based on cross-entropy (CE) loss to prevent the models from converging to local optima during the SSL process, thus further improving performance. Ablation experiments on CIFAR-10 with 50% symmetric noise and 40% asymmetric noise demonstrate that the two modules introduced in this paper improve the accuracy of the baseline (i.e., DivideMix) by 1.5% and 0.5%, respectively. Moreover, the experimental results on multiple benchmark datasets demonstrate that our proposed method effectively mitigates the impact of noisy labels and significantly enhances the performance of DNNs on noisy datasets. For instance, on the WebVision dataset, our method improves the top-1 accuracy by 0.7% and 2.4% compared to the baseline method. [ABSTRACT FROM AUTHOR]

Flagella are the main motility structure of Clostridioides difficile that affects the adhesion, colonization, and virulence of C. difficile in the human gastrointestinal tract. The FliL protein is a single transmembrane protein bound to the flagellar matrix. This study aimed to investigate the effect of the FliL encoding gene flagellar basal body-associated FliL family protein ( fliL ) on the phenotype of C. difficile . The fliL gene deletion mutant (Δ fliL ) and its corresponding complementary strains (: : fliL ) were constructed using allele-coupled exchange (ACE) and the standard molecular clone method. The differences in physiological properties such as growth profile, antibiotic sensitivity, pH resistance, motility, and spore production ability between the mutant and wild-type strains (CD630) were investigated. The Δ fliL mutant and the : : fliL complementary strain were successfully constructed. After comparing the phenotypes of strains CD630, Δ fliL , and : : fliL , the results showed that the growth rate and maximum biomass of Δ fliL mutant decreased than that of CD630. The Δ fliL mutant showed increased sensitivity to amoxicillin, ampicillin, and norfloxacin. Its sensitivity to kanamycin and tetracycline antibiotics decreased, and the antibiotic sensitivity partially returned to the level of CD630 strain in the : : fliL strain. Moreover, the motility was significantly reduced in the Δ fliL mutant. Interestingly, the motility of the : : fliL strain significantly increased even when compared to that of the CD630 strain. Furthermore, the pH tolerance of the Δ fliL mutant significantly increased or decreased at pH 5 or 9, respectively. Finally, the sporulation ability of Δ fliL mutant reduced considerably compared to the CD630 strain and recovered in the : : fliL strain. We conclude that the deletion of the fliL gene significantly reduced the swimming motility of C. difficile , suggesting that the fliL gene is essential for the motility of C. difficile . The fliL gene deletion significantly reduced spore production, cell growth rate, tolerance to different antibiotics, acidity, and alkalinity environments of C. difficile . These physiological characteristics are closely related to the survival advantage in the host intestine, which is correlated with its pathogenicity. Thus, we suggested that the function of the fliL gene is closely related to its motility, colonization, environmental tolerance, and spore production ability, which consequently affects the pathogenicity of C. difficile .