Abstract
Since Deep Convolutional Generative Adversarial Networks (DCGAN) was proposed, it has been perceived as a model with difficulty in training due to several factors. To solve this problem, ...dozens of optimization strategies were presented, but none of them was compared with the others. In this paper, the author chose three representative methods, namely one-label smoothing, the two Time-Scale Update Rule (TTUR), and the Earth-Mover Distance (EMD) or Wasserstein-1 to make a comparison of the optimization effect on the DCGAN model. To be specific, these three approaches were adopted respectively while using MNIST and Fashion-MNIST as datasets. One-side label smoothing was designed to prevent overconfidence in the model by adding a penalty term in the discriminator. TTUR was a simpler update strategy that could help the model find the stationary local Nash equilibrium under mild assumptions. EMD was an alternative loss function that enabled the model to distinguish the difference while the real distribution and generated distribution were not overlapped. Contrast experiments were conducted both vertically and horizontally. The author applied these three methods with the same dataset and the same method with different datasets in order to compare the time of the model collapse, the trend of loss in line graphs, and the impact of different datasets on results. Experimental results indicated that both one-label smoothing and TTUR postponed the model collapse while EMD completely get rid of it. Furthermore, generated images may lose texture information when using more complicated datasets.
In coherent direction of arrival (DOA) estimation, subspace-based methods suffer from performance deterioration because of the rank loss of the signal covariance matrix. A variety of spatial ...smoothing preprocessing techniques have been proposed for decorrelation, among which the enhanced spatial smoothing preprocessing (ESS) technique shows outstanding performance by exploiting the signal subspace. However, ESS is applied by squaring the matrix whose columns span the signal subspace (called signal matrix), which involves unnecessary computational loads. Besides, the smoothed covariance matrix after ESS is a linear combination of the sub-matrices of signal matrix, where the coefficients of the combination may undermine the decorrelation performance. In this context, a simplified spatial smoothing (SSS) technique is proposed for decorrelation by averaging the sub-matrices of the signal matrix directly, and avoids redundant operations in squared signal matrix. The proposed method is tested numerically in terms of the signal-to-noise ratio (SNR), the number of snapshots, angle separation, and the execution time. Simulation results show the improvement of the decorrelation performance, efficiency, and robustness with the proposed method in coherent scenarios, compared with the other spatial smoothing preprocessing based methods.
Multi-scale exposure fusion (MEF) is an efficient way to fuse differently exposed low dynamic range (LDR) images of a high dynamic range (HDR) scene into an information enriched LDR image. In this ...paper, a new MEF algorithm is proposed to merge the differently exposed LDR images by introducing novel content adaptive edge-preserving smoothing (CAS) pyramids for the weight maps of all the LDR images. With the proposed CAS pyramids, details in the brightest and darkest regions of the HDR scene are preserved better than existing MEF algorithms on top of the Gaussian pyramids and edge-preserving smoothing pyramids. Comparisons experimentally demonstrate the effectiveness of the proposed algorithm to nine state-of-the-art MEF algorithms from both subjective and objective points of view regardless the image sizes.
Energy is present in every touch we make in our modern life. However, with the increase of energy-consuming devices, there is a burden to meet this demand with a clean power generation. Although ...solar and wind are considered as clean and renewable energy resources, many issues engender with their higher penetration into electricity grids. However, when renewable energy resources are integrated with battery energy storage systems (BESS), more smoothed and easily dispatchable power can be obtained. This paper investigates the smoothing quality of the solar photovoltaic power output with the help of BESS using a couple of approaches such as low pass filtering (LPF), moving average (MA) filtering, Gaussian filter (GF) and Saviztky-Golay (S-G) filter. Obviously, the smoothed dispatchable power has been achieved with all mentioned methods, however, the performance of moving average and low pass filters is not acceptable comparatively especially when longer window size and time constant are used, which consequently deteriorates the performance of storage system. In contrast, the paper introduces using the Savitsky-Golay filter to reduce battery ramp rate and battery charging and discharging power while smoothing the solar power fluctuations. The simulation results depict the performance of the proposed smoothing filter and compare its performance against MA, LPF and GF.
The purpose of texture smoothing is to preserve the prominent structure in the image while smoothing the salient texture. However, the existing methods are difficult to achieve a satisfactory balance ...between filtering out salient textures and preserving weak edge structures and small structures. To this end, we propose a structure-preserving texture smoothing method via scale-aware bilateral total variation. First, the joint bilateral filter is introduced to construct the window bilateral variation, and combined with the window total variation, a regularizer called the bilateral total variation is formed, which accurately quantifies the characteristics of texture and structure, to finely smooth salient textures while preserving weak edge structures and small structures. Subsequently, we proposed a scale-aware scheme to make the proposed regularizer more powerful in preserving small structures and adopted an optimization scheme to convert the original non-convex optimization problem into a least squares regression problem. The effectiveness of the proposed regularizer is verified in the dataset. The experimental results demonstrate the superiority of the proposed method in texture smoothing and other applications compared to other state-of-the-art approaches.
This brief is devoted to the power-constrained low-thrust optimal control problem (OCP). First, a general model for a class of power constraints is proposed by defining a piecewise smooth function, ...which describes the discontinuities of the control system. Subsequently, a recursive smoothing function is proposed to smooth the generalized discontinuous terms leading to a family of smooth OCPs, whose solutions converge to the solution of the discontinuous power-constrained OCP. The distinguished feature is that the variations of the control bound, caused by the power constraints, are embedded into the smoothing function, and are retrieved gradually. Consequently, the constrained and unconstrained OCPs are seamlessly connected, allowing an easy start and convergence improvement of the continuation. Moreover, a homotopy is constructed to guarantee a smooth connection, providing a differential homotopy path-tracking process. Finally, by comparing with the existing methods, typical scenarios are simulated to demonstrate the convergence improvement and efficiency of the proposed generic homotopic smoothing method.
This is a second part of the research on AC optimal power flow being used in the lower level of the bilevel strategic bidding or investment models. As an example of a suitable upper-level problem, we ...observe a strategic bidding of energy storage and propose a novel formulation based on the smoothing technique. After presenting the idea and scope of our work, as well as the model itself and the solution algorithm in the companion paper (Part I), this paper presents a number of existing solution techniques and the proposed one based on smoothing the complementary conditions. The superiority of the proposed algorithm and smoothing techniques is demonstrated in terms of accuracy and computational tractability over multiple transmission networks of different sizes and different OPF models. The results indicate that the proposed approach outperforms all other options in both metrics by a significant margin. This is especially noticeable in the metric of accuracy where out of total 422 optimizations over 9 meshed networks the greatest AC OPF error is 0.023% that is further reduced to 3.3e-4% in the second iteration of our algorithm.
To achieve high resolution, orthogonal frequency division multiplex (OFDM) radars deploy two-dimensional multiple signal classification (2D-MUSIC) in the joint range-velocity estimator. However, it ...is obvious that both of the signal reconstruction and 2D smoothing affect the noise statistical distribution and virtual array aperture in the joint range-velocity estimator with 2D-MUSIC. Therefore, the conventional accuracy analysis methods for the MUSIC are no longer suitable. In this paper, we propose an estimation accuracy analysis for the joint range-velocity estimator with 2D-MUSIC. Firstly, we present that the noise statistical distribution of the signals being reconstructed and 2D smoothed has a 2-fold Hankel structure. Then, the closed-forms of the range and velocity estimation accuracies are derived, respectively, which both show that the estimation accuracies highly depend on the 2D smoothing and could be improved by properly choosing the 2D smoothing window. Further, we formulate the smoothing optimization and propose the quasi-optimal size of the 2D smoothing window. Finally, both theoretical analyses and simulations validate that the proposed smoothing optimization could significantly improve the estimation accuracies with a lower computational burden than the conventional smoothing.