The process of labeling samples costs time and resources but unlabeled samples are easier to obtain. Recently, graph-based deep semi-supervised learning (GDSSL) training a deep network using a small number of labeled ...
详细信息
ISBN:
(数字)9781728176871
ISBN:
(纸本)9781728176888
The process of labeling samples costs time and resources but unlabeled samples are easier to obtain. Recently, graph-based deep semi-supervised learning (GDSSL) training a deep network using a small number of labeled samples and the abundant unlabeled samples has been demonstrated to be promising on image classification tasks. These methods construct a graph to represent the structure of the input data (or hidden features). The successes of these GDSSL algorithms depend upon the structure of the similarity graph. However, existing GDSSL approaches construct the graph using predefined rules (such as knn graph) or fixed similarity measures (such as Gaussian kernel), which may limit the potential of GDSSL. In this paper, we move further in this direction to propose a novel end-to-end GDSSL approach which fully optimizes the whole graph without such limitations. To this end, we concatenate two neural networks (feature network and similarity network) together to learn the categorical label and semantic similarity, respectively, and train the networks with a new regularization term, the extended graph Laplacian, to minimize a unified objective function. Extensive experiments on several benchmark datasets demonstrate that our approach could outperform existing approaches on image classification. Furthermore, as a side-product, the similarity network could give faithful semantic similarity measure of samples, which is not possessed by other GDSSL approaches.
Random Fourier Features (RFF) demonstrate well-appreciated performance in kernel approximation for large-scale situations but restrict kernels to be stationary and positive definite. And for non-stationary kernels, th...
详细信息
TV constrained reconstruction could obtain perfect results from incomplete data, and has been applied to reduce metal artifact by assuming that the projection contaminated by metal is missing. In TV constrained recons...
详细信息
TV constrained reconstruction could obtain perfect results from incomplete data, and has been applied to reduce metal artifact by assuming that the projection contaminated by metal is missing. In TV constrained reconstruction, the selection of a proper step parameter for TV minimization procedure is a key point. However, this parameter is usually selected empirically, and it is a constant for all pixels in the whole image domain, regardless of the difference of missing projection quantity at different pixels. By analyzing the relationship between the missing projections and pixels position, a Weighted Total Variation (WTV) constrained reconstruction method is proposed to reduce metal artifact in this paper. For WTV constrained method, the parameters are no longer the same, but vary over image domain as the introduced information miss rate. The simulation results show that the proposed method is more effective than current TV constraint to reduce metal artifact. Moreover, WTV constrained method is extended to other incomplete projection problems.
This paper proposes a low energy-consuming cluster-based algorithm to protect data integrity and privacy named ILCCPDA, which can dynamically elect cluster head by LEACH clustering protocol and take the simple cluster...
详细信息
This paper proposes a low energy-consuming cluster-based algorithm to protect data integrity and privacy named ILCCPDA, which can dynamically elect cluster head by LEACH clustering protocol and take the simple cluster fusion approach to reduce the data transmission, thus reducing energy consumption. ILCCPDA can detect data integrity by adding homomorphic message authentication code and take the random key distribution mechanism for data encryption. It can solve the problem of the integrity, privacy and energy consumption in the wireless transmission of sensor data.
Regularization is a solution to solve the problem of unstable estimation of covariance matrix with a small sample set in Gaussian classifier. And multi-regularization parameters estimation is more difficult than singl...
详细信息
ISBN:
(纸本)9789898425843
Regularization is a solution to solve the problem of unstable estimation of covariance matrix with a small sample set in Gaussian classifier. And multi-regularization parameters estimation is more difficult than single parameter estimation. In this paper, KLIM-L covariance matrix estimation is derived theoretically based on MDL (minimum description length) principle for the small sample problem with high dimension. KLIM-L is a generalization of KLIM (Kullback-Leibler information measure) which considers the local difference in each dimension. Under the framework of MDL principle, multi-regularization parameters are selected by the criterion of minimization the KL divergence and estimated simply and directly by point estimation which is approximated by two-order Taylor expansion. It costs less computation time to estimate the multi-regularization parameters in KLIM-L than in RDA (regularized discriminant analysis) and in LOOC (leave-one-out covariance matrix estimate) where cross validation technique is adopted. And higher classification accuracy is achieved by the proposed KLIM-L estimator in experiment.
Radio frequency interference(RFI)is an important challenge in radio *** comes from various sources and increasingly impacts astronomical observation as telescopes become more *** this study,we propose a fast and effec...
详细信息
Radio frequency interference(RFI)is an important challenge in radio *** comes from various sources and increasingly impacts astronomical observation as telescopes become more *** this study,we propose a fast and effective method for removing RFI in pulsar *** use pseudo-inverse learning to train a single hidden layer auto-encoder(AE).We demonstrate that the AE can quickly learn the RFI signatures and then remove them from fast-sampled spectra,leaving real pulsar *** method has the advantage over traditional threshold-based filter method in that it does not completely remove contaminated channels,which could also contain useful astronomical information.
An improved MBNN (model-based neural network) was proposed to segment images. An image model obtained by the Markov random filed (MRF) was introduced into the MBNN. The MRF's parameters were estimated by modified ...
详细信息
An improved MBNN (model-based neural network) was proposed to segment images. An image model obtained by the Markov random filed (MRF) was introduced into the MBNN. The MRF's parameters were estimated by modified expectation-maximization (EM) algorithm. The technique of pre-assigning a class number was employed to decrease the computation burden. Therefore the task of image segmentation was implemented by the network. The experiment results show that it is feasible to apply the improved MBNN to image segmentation since a priori knowledge is excellently combined with local statistical correlation.
Bone scintigraphy is widely used to diagnose bone diseases. Accurate hotspot segmentation is a critical task for tumor metastasis diagnosis. In this paper, we propose an interactive approach to detect and extract hots...
详细信息
ISBN:
(纸本)9781479999897
Bone scintigraphy is widely used to diagnose bone diseases. Accurate hotspot segmentation is a critical task for tumor metastasis diagnosis. In this paper, we propose an interactive approach to detect and extract hotspots in thoracic region based on a new multiple instance learning (MIL) method called EM-MILBoost. We convert the segmentation problem to a multiple instance learning task by constructing positive and negative bags according to the input bounding box. In order to be robust against noisy input, we train a region-level hotspot classifier with EM-MILBoost and develop several segmentation strategies based on it. The experimental results demonstrate that our method outperforms other methods and is robust against various noisy input.
This paper presents a new feature extraction method for iris recognition. Since two dimensional complex wavelet transform (2D-CWT) does not only keep wavelet transform's properties of multiresolution decomposition...
详细信息
A closed form solution to the problem of segmenting multiple 3D motion models was proposed from straight-line optical flow. It introduced the multibody line optical flow constraint (MLOFC), a polynomial equation relat...
详细信息
A closed form solution to the problem of segmenting multiple 3D motion models was proposed from straight-line optical flow. It introduced the multibody line optical flow constraint (MLOFC), a polynomial equation relating motion models and line parameters. The motion models can be obtained analytically as the derivative of the MLOFC at the corresponding line measurement, without knowing the motion model associated with that line. Experiments on real and synthetic sequences were also presented.
暂无评论