In this paper, we propose a new neuralnetwork model termed semantic and episodic associative neuralnetwork (SEANN) for natural language processing. The SEANN can deal with both semantic memory and episodic memory by...
详细信息
ISBN:
(纸本)0780379527
In this paper, we propose a new neuralnetwork model termed semantic and episodic associative neuralnetwork (SEANN) for natural language processing. The SEANN can deal with both semantic memory and episodic memory by sentences represented in a form of a semantic network. In this model, both semantic memory and episodic memory are represented in triples-representation of concepts. Our model consists of concepts of sentences associative neuralnetwork (CSANN) and MAM using area representation. CSANN can recall sentences in a form of triples-representation, and MA M using area representation can recall plural triples-representations from a word. We have carried out computer experiments to confirm the validity of the SEANN for natural language processing. We have investigated that our model can recall plural semantic memories from one word, and can recall semantic memories concerning with episodic memory.
It is difficult to build a strict mathematical model for WEDM due to the complication of the machining process and the nonlinear relation between process parameters and process targets. The neuralnetwork is suited to...
详细信息
ISBN:
(纸本)9783642319648
It is difficult to build a strict mathematical model for WEDM due to the complication of the machining process and the nonlinear relation between process parameters and process targets. The neuralnetwork is suited to the modeling of complex system, because it has the functions of self-organized, self-learning and associative memory, and properties of distributed parallel type and high robustness. Therefore, this paper attempts to use the RBF neuralnetwork for the process modeling of WEDM.
The apparent dichotomy between symbolic AI processing and distributedneuralprocessing cannot be absolute, since neuralnetworks that capture essential features of human intelligence will also model some of the symbo...
详细信息
ISBN:
(纸本)0818642009
The apparent dichotomy between symbolic AI processing and distributedneuralprocessing cannot be absolute, since neuralnetworks that capture essential features of human intelligence will also model some of the symbolic processes of which humans are capable. Indeed, a primary goal of biological neuralnetwork research is to design systems that can self-organize intelligent symbolic processing capabilities. One such system is the ARTMAP family of neuralnetworks. Most if not all of the purported dichotomies between traditional artificial intelligence and neuralnetwork research dissolve within these systems. Although ARTMAP systems are neuralnetworks, they are also a type of self-organizing production system capable of hypothesis testing and memory search. They embody continuous and discrete, parallel and serial, and distributed and localized properties. Their symbols are compressed, often digital representations, yet they are formed and stabilized through a process of resonant binding that is distributed across the system. They are used to explain and predict data on both the psychological and the neurobiological levels, yet their unique combinations of computational properties are also rapidly finding their way into technology. They are capable of autonomously discovering rules about the environments to which they adapt, yet these rules are emergent properties of network dynamics rather than formal algorithmic statements.
We present a system for keyword spotting that, except for a front-end component for feature generation, it is entirely contained in a deep neuralnetwork (DNN) model trained "end-to-end" to predict the prese...
详细信息
ISBN:
(纸本)9781479981311
We present a system for keyword spotting that, except for a front-end component for feature generation, it is entirely contained in a deep neuralnetwork (DNN) model trained "end-to-end" to predict the presence of the keyword in a stream of audio. The main contributions of this work are, first, an efficient memoized neuralnetwork topology that aims at making better use of the parameters and associated computations in the DNN by holding a memory of previous activations distributed over the depth of the DNN. The second contribution is a method to train the DNN, end-to-end, to produce the keyword spotting score. This system significantly outperforms previous approaches both in terms of quality of detection as well as size and computation.
The aim of the paper is to present a parallel simulator that allows to use distributed algorithmic to develop artificial neuralnetworks. We have developed an efficient parallel simulator to implement artificial neura...
详细信息
ISBN:
(纸本)0780370449
The aim of the paper is to present a parallel simulator that allows to use distributed algorithmic to develop artificial neuralnetworks. We have developed an efficient parallel simulator to implement artificial neuralnetworks. This simulator makes use of the parallel properties of connectionist models to make an efficient parallel implementation onto general purpose shared memory MIMD computers. Therefore this simulator naturally leads to build (learning and generalising) neural algorithms that respect the large natural parallel aspects of these models.
A framework presenting a basic conceptual structure used to solve adaptive learning problems in soft real time applications is proposed. Its design consists of two supervised neuralnetworks running simultaneously. On...
详细信息
ISBN:
(纸本)9783642137686
A framework presenting a basic conceptual structure used to solve adaptive learning problems in soft real time applications is proposed. Its design consists of two supervised neuralnetworks running simultaneously. One is used for training data and the other is used for testing data. The accuracy of the classification is improved from the previous works by adding outpost vectors generated from prior samples. The testing function is able to test data continuously without being interrupted while the training function is being executed. The framework is designed for a parallel processing and/or a distributedprocessing environment due to the highly demanded processing power of the repetitive training process of the neuralnetwork.
In a two-node distributed multisensor system, some algorithms of track correlation can be transformed to a generalized classical assignment problem, and the generalized classical assignment is a combined optimization ...
详细信息
ISBN:
(纸本)9780780397361
In a two-node distributed multisensor system, some algorithms of track correlation can be transformed to a generalized classical assignment problem, and the generalized classical assignment is a combined optimization problem. The model of mean-field network is proposed in this paper to solve the problem. The experimental results illustrate that using the neuralnetwork to solve the generalized classical assignment problem not only makes the track association correct percent high, but also cannot increase the computing time exponentially with the number of targets.
Discrete Cosine Transformation (DCT) and Inverse Discrete Cosine Transformation (IDCT) are important parts of many image and video compression system. Unfortunately these operations are extremely computation-intensive...
详细信息
ISBN:
(纸本)0819442836
Discrete Cosine Transformation (DCT) and Inverse Discrete Cosine Transformation (IDCT) are important parts of many image and video compression system. Unfortunately these operations are extremely computation-intensive in a coding system, it consumes a large amount of resources for computation, especially in a real-time video coding system. In this article an efficient method for hardware based DCT/IDCT implementation is proposed. We combine the vector processing with parallel processing using distributed Arithmetic. At the same time the processing elements are pipelined to increase the processing speed and reduce the computation latency, which can also reduce the resource requirement and thus enhance the efficiency.
With the rapid development of the information age, the main body of public opinion information is no longer just structured data such as text and numbers, but the proportion of unstructured data such as audio and vide...
详细信息
ISBN:
(纸本)9781728183190
With the rapid development of the information age, the main body of public opinion information is no longer just structured data such as text and numbers, but the proportion of unstructured data such as audio and video has increased significantly. Therefore, the analysis and processing of unstructured public opinion data has become the key to deal with the problem of public opinion. Based on this, this paper designs an audio public opinion early warning model based on heterogeneous neuralnetwork. Heterogeneous neuralnetwork refers to a variety of different types of neuralnetworks to participate in the completion of specified tasks under the premise of data sharing, while retaining the autonomy of each neuralnetwork. In this paper, the audio data is converted into a spectrum as an input feature. The convolution neuralnetwork can obtain the feature representation of the audio in the spectrum graph, which is used as the input of the recurrent neuralnetwork, and then the features related to the text representation are obtained. In this paper, Bert is used to modify the semantic results combined with public opinion domain knowledge, and finally the results are introduced into the shallow neuralnetwork Fasttext for public opinion early warning processing. In this paper, experiments are carried out based on the open source voice data set ST-CMDS-20170001-1 Chinese mandarin corpus and Pytorch platform. The results show that this model can effectively analyze and process public opinion of unstructured audio and video data.
This paper proposes the use of massively parallel learning networks for classifying signals from electromagnetic transducers used in nondestructive evaluation. A description of the application problem is given. One of...
详细信息
ISBN:
(纸本)0852963882
This paper proposes the use of massively parallel learning networks for classifying signals from electromagnetic transducers used in nondestructive evaluation. A description of the application problem is given. One of the major contributions of the paper lies in the development of a distributedprocessingnetwork for preprocessing the transducer signal. Preprocessing is required for achieving invariance under rotation, translation and scaling of the signal. Issues relating to implementation of the preprocessingnetwork are presented. The complete architecture learning algorithm are described. Results of implementing the network as well as a discussion on the performance of the network are presented.
暂无评论