-In this paper, a motor control algorithm for performing a mode change without an integrated starter generator (ISG) is suggested for the automatic transmission-based hybrid electric vehicle (HEV). Dynamic models of the HEV powertrains such as engine, motor, and mode clutch are derived for the transient state during the mode change, and the HEV performance simulator is developed. Using the HEV performance bench tester, the characteristics of the mode clutch torque are measured and the motor torque required for the mode clutch synchronization is determined. Based on the dynamic models and the mode clutch torque, a motor torque control algorithm is presented for mode changes, and motor control without the ISG is investigated and compared with the existing ISG control.

No abstract

Due to the risk of data leakage while training deep learning models in a shared environment, we propose a new privacy-preserving deep learning(PPDL) method using a structural image deidentification approach for object classification. The proposed structural image de-identification approach is designed based on the fact that the degree of structural distortion of an image object has the greatest impact on human's perceptual system. Thus, by modifying only the structural parts of the original one using order preserving encryption(OPE), the proposed structural image de-identification approach decreases only the recognition rate by human. From the experimental results using different standard datasets, we show that the object classification accuracy of the proposed structural image de-identification method is almost the same as the deep learning performance for non-encrypted images, without revealing the original image contents including sensitive information. Also, by handling the trade-off between object classification accuracy and privacy protection for the de-identified image, we experimentally find the optimal size of input image for the proposed structural image de-identification approach.

Infectious diseases and pandemics, including the COVID-19 pandemic, have a huge economic impact on cities. However, few studies examine the economic resilience of small-scale regions within cities. Thus, this study derives neighborhoods with high economic resilience in a pandemic situation and reveals their urban characteristics. It evaluates economic resilience by analyzing changes in the amount of credit card payments in the neighborhood and classifying the types of neighborhoods therefrom. The study conducted the ANOVA, Kruskal–Wallis, and post hoc tests to analyze the difference in urban characteristics between neighborhood types. Accordingly, three neighborhood types emerged from the analysis: high-resilient neighborhood, low-resilient neighborhood, and neighborhood that benefited from the pandemic. The high-resilient neighborhood is a low-density residential area where many elderly people live. Neighborhoods that benefited are residential areas mainly located in high-density apartments where many families of parents and children live. The low-resilient neighborhood is an area with many young people and small households, many studio-type small houses, and a high degree of land-use mix.

Adversarial examples are human-imperceptible perturbations to inputs to machine learning models. While attacking machine learning models, adversarial examples cause the model to make a false positive or a false negative. So far, two representative defense architectures have shown a significant effect: (1) model retraining architecture; and (2) input transformation architecture. However, previous defense methods belonging to these two architectures do not produce good outputs for every input, i.e., adversarial examples and legitimate inputs. Specifically, model retraining methods generate false negatives for unknown adversarial examples, and input transformation methods generate false positives for legitimate inputs. To produce good-enough outputs for every input, we propose and evaluate a new input transformation architecture based on twostep input transformation. To solve the limitations of the previous two defense methods, we intend to answer the following question: How to maintain the performance of Deep Neural Network (DNN) models for legitimate inputs while providing good robustness against various adversarial examples? From the evaluation results under various conditions, we show that the proposed two-step input transformation architecture provides good robustness to DNN models against state-of-the-art adversarial perturbations, while maintaining the high accuracy even for legitimate inputs.