Neural computation could benefit from the heterogeneity of neurons to achieve energy efficiency. Beyond a single neuron level, adaptation to biologically important signals should also make functional columns heterogen...
详细信息
Neural computation could benefit from the heterogeneity of neurons to achieve energy efficiency. Beyond a single neuron level, adaptation to biologically important signals should also make functional columns heterogeneous. In the present study, we test a hypothesis that variability of neural response depends on tonotopic columns in the primary auditory cortex (A1) of rats. Mutual information (MI) was estimated from multi-unit responses in A1 of anesthetized rats, to quantify how spike count (SC) and the first spike latency (FSL) carried information about frequency and intensity of test tones. Consequently, for both SC and FSL, we found best frequency (BF)-dependent MI distributions with wide variances in high BF regions. These MI distributions were caused by BF-dependence of the amount of information that neurons conveyed, i.e., total entropy, rather than the transmission efficiency. In addition, the relationship between the transmission efficiency and the total entropy differentiated SC encoding and FSL encoding, suggesting that SC encoding and FSL encoding are not redundant but each plays a different role in intensity encoding. These results provide compelling evidence that BF columns are heterogeneous. Such heterogeneity of columns may make the global computation in A1 more efficient. Thus, the efficient coding in the neural system could be achieved by multiple-scale heterogeneity. (c) 2012 IBRO. Published by Elsevier Ltd. All rights reserved.
In the field of literature, there is an established set of techniques that have been successfully leveraged in the statistical analysis of literary style, most often to answer questions of authenticity and attribution...
详细信息
In the field of literature, there is an established set of techniques that have been successfully leveraged in the statistical analysis of literary style, most often to answer questions of authenticity and attribution. With the digitization of huge troves of art images come significant opportunities for the development of statistical techniques for the analysis of artistic style. In this article, we suggest that the progress made and statistical techniques developed in understanding visual processing as it relates to natural scenes can serve as a useful model and inspiration for visual stylometric analysis. (C) 2011 Wiley Periodicals, Inc.
In this paper, we address the problem of hallucinating a high resolution face given a low resolution input face. The problem is approached through sparse coding. To exploit the facial structure, Non-negative Matrix Fa...
详细信息
ISBN:
(纸本)9781424417650
In this paper, we address the problem of hallucinating a high resolution face given a low resolution input face. The problem is approached through sparse coding. To exploit the facial structure, Non-negative Matrix Factorization (NMF) [1] is first employed to learn a localized part-based subspace. This subspace is effective for super-resolving the incoming low resolution face under reconstruction constraints. To further enhance the detailed facial information, we propose a local patch method based on sparse representation with respect to coupled overcomplete patch dictionaries, which can be fast solved through linear programming. Experiments demonstrate that our approach can hallucinate high quality super-resolution faces.
A new natural image denoising method using it modified sparse coding (SC) algorithm proposed by us was discussed in this paper. This SC algorithm exploited the maximum Kurtosis as the maximizing sparse measure Criteri...
详细信息
ISBN:
(纸本)9783540877332
A new natural image denoising method using it modified sparse coding (SC) algorithm proposed by us was discussed in this paper. This SC algorithm exploited the maximum Kurtosis as the maximizing sparse measure Criterion at one time, a fixed variance term of sparse coefficients is used to yield a fixed information capacity. On the other hand, in order to improve the convergence speed. we use a determinative basis function as the initialization feature basis function of our sparse coding algorithm instead of using a random initialization matrix. This denoising method is evaluated by values of the normalized mean squared error (NMSE) and signal to noise ratio (NSNR). Compared with other denoising methods, the simulation results show (hat our SC shrinkage technique is indeed effective.
A novel dictionary learning design, driven by the Human Visual System (HVS) perception characteristic, for scalable representation of natural images is proposed. It builds upon the K-SVD algorithm, which learns non-sc...
详细信息
ISBN:
(纸本)9781467350518
A novel dictionary learning design, driven by the Human Visual System (HVS) perception characteristic, for scalable representation of natural images is proposed. It builds upon the K-SVD algorithm, which learns non-scalable dictionaries for natural images. We introduce regularization over the K-SVD dictionary atom update stage, enabling scalable sparse image reconstruction. Mainly, emphasis is on the dictionary's low and high spatial frequency components. Experimental results demonstrate the practicality of the proposed scheme for effective scalable sparse recovery of dynamic data changing over time (e. g., video). For the aforementioned purpose the proposed method outperforms the conventional K-SVD algorithm on average by 10.8[dB].
Given the increasing number of mobile platforms, a key technical challenge is how to provide an optimal photo browsing experience given the limited screen size available on mobile devices. This paper proposes a novel ...
详细信息
ISBN:
(纸本)9781467325882;9781467325875
Given the increasing number of mobile platforms, a key technical challenge is how to provide an optimal photo browsing experience given the limited screen size available on mobile devices. This paper proposes a novel technique for intelligent mobile image categorization on mobile platform to reduce computation complexity based on cloud computing. In this technique, captured images are analyzed to detect visual salient area, which is then classified in real-time using sparse representation. Mathematically, the derived algorithm regards the salient regions as the dictionary in sparse representation, and selects the salient regions that minimize the residual output error iteratively, thus the resulting regions have a direct correspondence to the performance requirements of the given problem. Experimental results obtained using extensive datasets captured under uncontrolled conditions show the proposed system effectively manages mobile images using sparse representation on cloud computing.
sparse coding learns its basis non-linearly, but the basis elements are still linearly combined to form an image. Is this linear combination of basis elements a good model for natural images? We here use a non-linear ...
详细信息
ISBN:
(纸本)9780819489388
sparse coding learns its basis non-linearly, but the basis elements are still linearly combined to form an image. Is this linear combination of basis elements a good model for natural images? We here use a non-linear synthesis rule, such that at each location in the image the point-wise maximum over all basis elements is used to synthesize the image. We present algorithms for image approximation and basis learning using this synthesis rule. With these algorithms we explore the the pixel-wise maximum over the basis elements as an alternative image model and thus contribute to the problem of finding a proper representation of natural images.
This paper presents a new image denoising method based on sparse reconstruction by dictionary learning and collaborative filtering. First, we form an affinity net, in which a node represents an image patch, for the gi...
详细信息
ISBN:
(纸本)9781467300469
This paper presents a new image denoising method based on sparse reconstruction by dictionary learning and collaborative filtering. First, we form an affinity net, in which a node represents an image patch, for the given image by clustering similar patches. For each cluster, we learn an undercomplete dictionary and represent clusters nodes by imposing sparsity inducing norm as a combination of few atoms. Depending on its affinity to other nodes, a single node could be present in multiple clusters making the clusters overlapping. This enables a single global estimation for each filtered pixel to be obtained by collaboratively aggregating its reconstructed patches in the corresponding clusters. Extensive experimental results demonstrate superior performance for additive noise removal without requiring the correct noise variance.
A novel image reconstruction method for natural images using a modified sparse coding (SC) algorithm is proposed by us. This SC algorithm exploited the maximum Kurtosis as the maximizing sparse measure criterion, and ...
详细信息
ISBN:
(纸本)9783540874409
A novel image reconstruction method for natural images using a modified sparse coding (SC) algorithm is proposed by us. This SC algorithm exploited the maximum Kurtosis as the maximizing sparse measure criterion, and a fixed variance term of sparse coefficients is used to yield a fixed information capacity. The experimental results show that using our algorithm, the natural images' feature basis vectors can be successfully extracted. Furthermore, compared with the standard SC method, the experimental results show that our algorithm is indeed efficient and effective in performing image reconstruction task.
This work considers the problem of learning an incoherent dictionary that is both adapted to a set of training data and incoherent so that existing sparse approximation algorithms can recover the sparsest representati...
详细信息
ISBN:
(纸本)9781467300469
This work considers the problem of learning an incoherent dictionary that is both adapted to a set of training data and incoherent so that existing sparse approximation algorithms can recover the sparsest representation. A new decorrelation method is presented that computes a fixed coherence dictionary close to a given dictionary. That step iterates pairwise decorrelations of atoms in the dictionary. Dictionary learning is then performed by adding this decorrelation method as an intermediate step in the K-SVD learning algorithm. The proposed algorithm INK-SVD is tested on musical data and compared to another existing decorrelation method. INK-SVD can compute a dictionary that approximates the training data as well as K-SVD while decreasing the coherence from 0.6 to 0.2.
暂无评论