Search

Most studies on convolutional Neural Network (CNN) based image processing have proposed networks that can be optimized for a single level. Here, the term "level" refers to the specific objective defined for each task, such as the degree of noise in denoising tasks. Hence, they underperform on other levels and must be retrained to deliver optimal performance. Using multiple models to cover multiple levels involves very high computational costs. To solve these problems, recent approaches train the networks on two different levels and propose their own modulation methods to enable the arbitrary intermediate levels. However, many of them 1) have difficulty adapting from one level to the other, 2) suffer from unintended artifacts in the intermediate levels, or 3) require large memory and computational cost. In this paper, we propose a novel framework using Filter Transition Network (FTN), which is a non-linear module that easily adapts to new levels, is regularized to prevent undesirable side-effects, and extremely lightweight being a data-independent module. Additionally, for stable learning of FTN, we newly propose a method to initialize nonlinear CNNs with identity mappings. Extensive results for various image processing tasks indicate that the performance of FTN is stable regarding adaptation and modulation and is comparable to that of the other heavy frameworks. [ABSTRACT FROM AUTHOR]

This paper presents a dynamic voltage restorer to solve the voltage sag problem that causes the most serious economic loss among various grid accidents. The proposed method, which uses a direct AC–AC converter, consists of a simple controller unlike methods using a voltage source inverter that is controlled using complex operations. Unlike the energy storage type, there is no additional energy storage device, which reduces system cost. In addition, unlike the back-to-back type, energy conversion loss is reduced by one stage conversion. A switching cell-structured direct AC–AC converter that solves the rectification problem without a separate sensor is a reliable structure that reduces the volumes of the passive filters by increasing the switching frequency. In addition, using interphase voltages, it can take a wide range of voltage compensations to compensate for most of the voltage sag in various situations. In this paper, the background of the study is revealed and the operating principles and control schemes of the proposed system are analyzed. Then, the validity of this study is confirmed through simulation and experiment results. [ABSTRACT FROM AUTHOR]

Summary: Circadian rhythms in mammals are generated by a transcriptional negative feedback loop that is driven primarily by oscillations of PER and CRY, which inhibit their own transcriptional activators, CLOCK and BMAL1. Current models posit that CRY is the dominant repressor, while PER may play an accessory role. In this study, however, constitutive expression of PER, and not CRY1, severely disrupted the clock in fibroblasts and liver. Furthermore, constitutive expression of PER2 in the brain and SCN of transgenic mice caused a complete loss of behavioral circadian rhythms in a conditional and reversible manner. These results demonstrate that rhythmic levels of PER2, rather than CRY1, are critical for circadian oscillations in cells and in the intact organism. Our biochemical evidence supports an elegant mechanism for the disparity: PER2 directly and rhythmically binds to CLOCK:BMAL1, while CRY only interacts indirectly; PER2 bridges CRY and CLOCK:BMAL1 to drive the circadian negative feedback loop. [Copyright &y& Elsevier]