Winner-Take-All (WTA) or K-Winner-Take-All (KWTA) networks have been frequently used as the basic building blocks of complex neural networks. This paper introduces a new selection rule for network connections that implements stable KWTA networks. To widen the applications of WTA networks, a new class of WTA networks is proposed, and their efficient design methods are presented. We demonstrate the properties of the generalized class of WTA networks, through three application examples. 相似文献
Artificial Intelligence Review - Visual object tracking has become one of the most active research topics in computer vision, and it has been applied in several commercial... 相似文献
Achieving high processing quality for chemical mechanical planarization (CMP) in semiconductor manufacturing is difficult due to the distinct process variations associated with this method, such as drift and shift. Run-to-run control aims to maintain the targeted process quality by reducing the effect of process variations. The goal of controller learning is to infer an underlying output–input reverse mapping based on input–output samples considering the process variations. Existing controllers learn reverse mapping by minimizing the total mapping error for sample data. However, this approach often fails to generate inputs for unseen target outputs because conditional input distributions on target outputs are not captured in the learning. In this study, we propose a controller based on a least squares generative adversarial network (LSGAN) that can capture the input distributions. GANs are deep-learning architectures composed of two neural nets: a generator and a discriminator. In the proposed model, the generator attempts to produce fake input distributions that are similar to the real input distributions considering the process variation features extracted using convolutional layers, while the discriminator attempts to detect the fake distributions. Competition in this game drives both networks to improve their performance until the generated input distributions are indistinguishable from the real distributions. An experiment using the data obtained from a work-site CMP tool verified that the proposed model outperformed the comparison models in terms of control accuracy and computation time.
The squash-and-stretch describes the rigidity of the character. This effect is the most important technique in traditional cartoon animation. In this paper, we introduce a method that applies the squash-and-stretch effect to character motion. Our method exaggerates the motion by sequentially applying the spatial exaggeration technique and the temporal exaggeration technique. The spatial exaggeration technique globally deforms the pose in order to make the squashed or stretched pose by modeling it as a covariance matrix of joint positions. Then, the temporal exaggeration technique computes a time-warping function for each joint, and applies it to the position of the joint allowing the character to stretch its links appropriately. The motion stylized by our method is a sequence of squashed and stretched poses with stretching limbs. By performing a user survey, we prove that the motion created using our method is similar to that used in 2D cartoon animation and is funnier than the original motion for human observers who are familiar with 2D cartoon animation. 相似文献
Due to the advancement of wireless internet and mobile positioning technology, the application of location-based services (LBSs) has become popular for mobile users. Since users have to send their exact locations to obtain the service, it may lead to several privacy threats. To solve this problem, a cloaking method has been proposed to blur users’ exact locations into a cloaked spatial region with a required privacy threshold (k). With the cloaked region, an LBS server can carry out a k-nearest neighbor (k-NN) search algorithm. Some recent studies have proposed methods to search k-nearest POIs while protecting a user’s privacy. However, they have at least one major problem, such as inefficiency on query processing or low precision of retrieved result. To resolve these problems, in this paper, we propose a novel k-NN query processing algorithm for a cloaking region to satisfy both requirements of fast query processing time and high precision of the retrieved result. To achieve fast query processing time, we propose a new pruning technique based on a 2D-coodinate scheme. In addition, we make use of a Voronoi diagram for retrieving the nearest POIs efficiently. To satisfy the requirement of high precision of the retrieved result, we guarantee that our k-NN query processing algorithm always contains the exact set of k nearest neighbors. Our performance analysis shows that our algorithm achieves better performance in terms of query processing time and the number of candidate POIs compared with other algorithms. 相似文献
In this paper, we propose an advanced partial encryption of watermarking and scrambling using the magnitude information of Modifed Discrete Cosine Transform (MDCT). In MPEG-1/Audio Layer III (MP3), the magnitude and phase information of modified discrete cosine transform (MDCT) coefficients is encrypted. The proposed method uses both watermarking and scrambling, and aims at protecting the contents against eavesdropping and moreover against illegal mass distribution after descrambled. Experimental results show that the proposed method can achieve higher security and less computational complexity by reusing the MDCT coefficients obtained in MP3. 相似文献
A general optimization formulation for transition walking prediction using 3D skeletal model is presented. The formulation
is based on a previously presented one-step walking formulation (Xiang et al., Int J Numer Methods Eng 79:667–695, 2009b). Two basic transitions are studied: walk-to-stand and slow-to-fast walk. The slow-to-fast transition is used to connect
slow walk to fast walk by using a step-to-step transition formulation. In addition, the speed effects on the walk-to-stand
motion are investigated. The joint torques and ground reaction forces (GRF) are recovered and analyzed from the simulation.
For slow-to-fast walk transition, the predicted ground reaction forces in step transition is even larger than that of the
fast walk. The model shows good correlation with the experimental data for the lower extremities except for the standing ankle
profile. The optimal solution of transition simulation is obtained in a few minutes by using predictive dynamics method. 相似文献
The upper and lower limits of the electrostrictive constants, dielectric permittivities, spontaneous polarizations, and piezoelectric coefficients were calculated for ceramic PbTiO(3) from theoretical single-crystal constants. Experimental ceramic data fall between these upper and lower limits. The large piezoelectric anisotropy d(33)/d(31) of ceramic PbTiO(3 ) was shown to be related to the single-crystal PbTiO(3) electrostrictive anisotropies Q(11)/Q(12 ) and Q(44)/Q(12). The possibility of a change in sign of the ceramic d(31) coefficient due to a slight variation in the single-crystal electrostrictive anisotropies was discussed. The single-crystal and predicted ceramic hydrostatic electrostrictive constants were found to be equal. Using this result the ceramic hydrostatic g(h ) coefficient is always smaller than the single-crystal g (h), but the ceramic hydrostatic d(h) coefficient can be either larger or smaller than the single-crystal d(h) depending on the dielectric anisotropy (epsilon (11)/epsilon(33)) of the single-crystal. 相似文献