Deep neuralnetworks (DNNs) have been vastly and successfully employed in various artificial intelligence and machine learning applications (e.g., imageprocessing and natural language processing). As DNNs become deep...
详细信息
Deep neuralnetworks (DNNs) have been vastly and successfully employed in various artificial intelligence and machine learning applications (e.g., imageprocessing and natural language processing). As DNNs become deeper and enclose more filters per layer, they incur high computational costs and large memory consumption to preserve their large number of parameters. Moreover, present processing platforms (e.g., CPU, GPU, and FPGA) have not enough internal memory, and hence external memory storage is needed. Hence deploying DNNs on mobile applications is difficult, considering the limited storage space, computation power, energy supply, and real-time processing requirements. In this work, using a method based on tensor decomposition, network parameters were compressed, thereby reducing access to external memory. This compression method decomposes the network layers' weight tensor into a limited number of principal vectors such that (i) almost all the initial parameters can be retrieved, (ii) the network structure did not change, and (iii) the network quality after reproducing the parameters was almost similar to the original network in terms of detection accuracy. To optimize the realization of this method on FPGA, the tensor decomposition algorithm was modified while its convergence was not affected, and the reproduction of network parameters on FPGA was straightforward. The proposed algorithm reduced the parameters of ResNet50, VGG16, and VGG19 networks trained with Cifar10 and Cifar100 by almost 10 times. (C)& nbsp;2022 Elsevier Ltd. All rights reserved.
It is proved analytically, whenever the input-output mapping of a one-layered hard-limited perceptron satisfies a positive, linear independency (PLI) condition, the connection matrix A to meet this mapping can be obta...
详细信息
ISBN:
(纸本)0819424412
It is proved analytically, whenever the input-output mapping of a one-layered hard-limited perceptron satisfies a positive, linear independency (PLI) condition, the connection matrix A to meet this mapping can be obtained noniteratively in one step from an algebraic matrix equation containing an NxM input matrix U. Each column of U is a given standard pattern vector, and there are M standard patterns to be classified It is also analytically proved that sorting out all nonsingular sub-matrices U-k in U can be used as an automatic feature extraction process in this noniterative-learning system. This paper reports the theoretical derivation and the design and experiments of a superfast-learning, optimally-robust, neural network pattern recognition system utilizing this novel feature extraction process. An unedited video movie showing the speed of learning and the robustness in recognition of this novel pattern recognition system will be demonstrated in life. Comparison to other neural network pattern recognition systems will be discussed.
Besides the variety of fonts, character recognition systems for industrial world are confronted with specific problems like : the variety of support (metal, wood, paper, ceramics...) as well as the variety of marking ...
详细信息
ISBN:
(纸本)0819424412
Besides the variety of fonts, character recognition systems for industrial world are confronted with specific problems like : the variety of support (metal, wood, paper, ceramics...) as well as the variety of marking (printing, engraving,...) and conditions of lighting. We present a system that is able to solve a part of this problem. It implements a collaboration between two neuralnetworks. The first network specialized in vision allows the system to extract the character from an image. Besides this capability, we have equipped our system with characteristics allowing it to obtain an invariant model from presented character, Thus, whatever the position, the size and the orientation of the character during the capture are, the model presented to the input of the second network will be identical. The second network, thanks to a learning phase, permits to obtain a character recognition system independent of the type of fonts used. Furthermore, its capabilities of generalization permit to recognize degraded and/or distorted characters, A feedback loop between the two networks permits to the first one to modify the quality of vision, The cooperation between these two networks allows to recognize characters whatever are the support and the marking.
Adaptive Solutions' CNAPS architecture is a parallel array of digital processors. This design features a Single-Instruction Multiple-Data (SIMD) stream architecture. The architecture is designed to execute on- chi...
详细信息
ISBN:
(纸本)0819405515
Adaptive Solutions' CNAPS architecture is a parallel array of digital processors. This design features a Single-Instruction Multiple-Data (SIMD) stream architecture. The architecture is designed to execute on- chip learning for artificialneural Network (ANN) algorithms with unprecedented performance. ANNs have shown impressive results for solving difficult imageprocessing tasks. However, current hardware prevents many ANN solutions from being effective products. The CNAPS architecture will provide the computational power to allow real time ANN applications. Because of the high parallelism of the architecture,it is also ideal for digital imageprocessing tasks. This architecture will allow high performance applications that combine conventional imageprocessing methods and ANNs on the same system. This paper gives a brief introduction to the CNAPS architecture, and gives the system performance on implementation of neural network algorithms, and conventional imageprocessing algorithms such as convolution, and 2D Fourier transforms.
This paper presents a bibliography of nearly 1700 references related to computer vision and image analysis, arranged by subject matter. The topics covered include computational techniques;feature detection and segment...
This paper presents a bibliography of nearly 1700 references related to computer vision and image analysis, arranged by subject matter. The topics covered include computational techniques;feature detection and segmentation;image and scene analysis;two-dimensional shape;pattern;color and texture;matching and stereo;three-dimensional recovery and analysis;three-dimensional shape;and motion. A few references are also given on related topics, including geometry and graphics, compression and processing, sensors and optics, visual perception, neuralnetworks, artificial intelligence and pattern recognition, as well as on applications. (C) 1998 Academic Press.
The ML Parameter Estimation and the neural Network based methods for classifying the textures are compared in this paper. The comparison is based on the correct classification percentage. Certain constraints have been...
详细信息
ISBN:
(纸本)0819424412
The ML Parameter Estimation and the neural Network based methods for classifying the textures are compared in this paper. The comparison is based on the correct classification percentage. Certain constraints have been imposed on the classifiers which are using the same sample size, same number of features and same number of training and test feature vectors for both the classifiers. The classifiers use the energy of the dominant channels of a tree-structured wavelet transform as features. Experiments are performed with textures from the Brodatz album. All the textured images are of size 256 X 256 pixels with 256 gray levels. Selection of best feature set has been arrived at using the ''leave one out'' approach. The results indicate that both the classifiers give comparable performance. However, the governing factors for their choice is the number of training samples, number of features, and the computational complexity for both the classifiers, and the size of the network, in specific, for the neural network.
Interesting perspectives in imageprocessing with cellular neuralnetworks can be emphasized from an investigation into the internal states dynamics of the model. Most of the cellular neuralnetworks design methods in...
详细信息
ISBN:
(纸本)3540660682
Interesting perspectives in imageprocessing with cellular neuralnetworks can be emphasized from an investigation into the internal states dynamics of the model. Most of the cellular neuralnetworks design methods intend to control internal states dynamics in order to pet a straight processing result. The present one involves some kind of internal states preprocessing so as to finally achieve processing otherwise unrealizable. applications of this principle to the building of complex processing schemes, gray level preserving segmentation and selective brightness variation are presented.
artificial retina chips which can simultaneously sense and process the real world images are described. Device concept, structure, fundamental performance, operation principle, processing functions are described. Appl...
详细信息
ISBN:
(纸本)0780341236
artificial retina chips which can simultaneously sense and process the real world images are described. Device concept, structure, fundamental performance, operation principle, processing functions are described. applications including the interactive games by gesture-input are also introduced.
Current efforts to perform automatic galaxy classification using artificialneural network image classifiers are reviewed. For both digitized photographic Schmidt plate data and newly obtained WFPC2 imagery from the H...
详细信息
ISBN:
(纸本)0819425869
Current efforts to perform automatic galaxy classification using artificialneural network image classifiers are reviewed. For both digitized photographic Schmidt plate data and newly obtained WFPC2 imagery from the Hubble Space Telescope, a variety of two-dimensional photometric parameter spaces produce a segregation of Hubble types. Through the use of hidden node layers, a neural network is capable of mapping complicated, highly nonlinear data space. This powerful technique is used to map a multivariate photometric parameter space to the revised Hubble system of galaxy classification.
This paper extends a recent and very appealing approach of computational learning to the field of image analysis. Recent works have demonstrated that the implementation of artificialneuralnetworks (ANN) could be sim...
详细信息
ISBN:
(纸本)9783642217371;9783642217388
This paper extends a recent and very appealing approach of computational learning to the field of image analysis. Recent works have demonstrated that the implementation of artificialneuralnetworks (ANN) could be simplified by using a large amount of neurons with random weights. Only the output weights are adapted, with a single linear regression. Supervised learning is very fast and efficient. To adapt this approach to image analysis, the novelty is to initialize weights, not as independent random variables, but as Gaussian functions with only a few random parameters. This creates smooth random receptive fields in the image space. These image Receptive Fields - neuralnetworks (IRF-NN) show remarkable performances for recognition applications, with extremely fast learning, and can be applied directly to images without pre-processing.
暂无评论