Container environments permeate all areas of computing, such as HPC, since they are lightweight, efficient, and ease the deployment of software. However, due to the shared host kernel, their isolation is considered to...
详细信息
ISBN:
(纸本)9781728165820
Container environments permeate all areas of computing, such as HPC, since they are lightweight, efficient, and ease the deployment of software. However, due to the shared host kernel, their isolation is considered to be weak, so additional protection mechanisms are needed. This paper shows that neuralnetworks can be used to do anomaly detection by observing the behavior of containers through system call data. In more detail the detection of anomalies in file and directory paths used by system calls is evaluated to show their advantages and drawbacks.
We perform a theoretical analysis comparing the scalability of data versus model parallelism, applied to the distributed training of deep convolutional neuralnetworks (CNNs), along live axes: batch size, node (floati...
详细信息
ISBN:
(纸本)9781665414555
We perform a theoretical analysis comparing the scalability of data versus model parallelism, applied to the distributed training of deep convolutional neuralnetworks (CNNs), along live axes: batch size, node (floating-point) arithmetic performance, node memory bandwidth, network link bandwidth, and cluster dimension. Our study relies on analytical performance models that can he configured to reproduce the components and organization of the CNN model as well as the hardware configuration of the target distributed platform. In addition, we provide evidence of the accuracy of the analytical models by performing a validation against a Python library for distributed deep learning training.
This paper proposes wireless sensor networks as a parallel and distributed computing platform for neurocomputing. The proposal entails leveraging the existing wireless sensor networks technology to serve as a hardware...
详细信息
ISBN:
(纸本)9781467361293;9781467361286
This paper proposes wireless sensor networks as a parallel and distributed computing platform for neurocomputing. The proposal entails leveraging the existing wireless sensor networks technology to serve as a hardware-software platform to implement and realize artificial neuralnetwork algorithms in fully parallel and distributed computation mode. The study describes the proposed parallel and distributed neurocomputing architecture, which is named as WSN-ANN, and its use as a hardware platform on a case study. A Hopfield neuralnetwork, which is configured to solve the minimum weakly connected dominating set problem, is embedded into a wireless sensor network. Simulation study results indicate that the proposed computing platform based on wireless sensor networks, WSN-ANN, is feasible and promising to serve as a parallel and distributed neurocomputer.
This paper presents results of using deep learning neuralnetwork techniques to map words, represented at the levels of letters, into semantic, distributed vector representations. We study and compare algorithms that ...
详细信息
ISBN:
(纸本)9781509060146
This paper presents results of using deep learning neuralnetwork techniques to map words, represented at the levels of letters, into semantic, distributed vector representations. We study and compare algorithms that have been proposed as models for human orthographic and morphological processing, such as the two layers symbolic network (AKA Naive Discriminatory Learning network) proposed by Baayen et al [1], and the dual-route approach presented by Grainger and Ziegler [2]. In addition, we study the effect of representing letters as one-hot vectors or via distributed vector representations, much like the natural language processing field has done for words. Experiment results show a better performance for the dual-route algorithm using distributed letter representations, both in terms of accuracy and speed. Our results are obtained with training sets in the tens of thousands of words, as opposed to the hundreds of thousands of words used elsewhere in the literature. These results point to the feasibility of taking advantage of morpho-orthographic patterns in order to assist with word processing in natural language processing tasks.
The goal of the provider with respect to dynamic resource allocation in cloud computing is to maintain application performance according to service level agreements while reducing electrical power costs. To achieve th...
详细信息
ISBN:
(纸本)9781479927289
The goal of the provider with respect to dynamic resource allocation in cloud computing is to maintain application performance according to service level agreements while reducing electrical power costs. To achieve this goal, we present a resource manager that optimizes a utility function expressing the trade-off between the conflicting objectives of maintaining application performance and reducing power costs. It is based on an artificial neuralnetwork (ANN) to find the best resource allocation to virtual machines that optimizes the utility function. To provide support for a potentially large number of virtual machines, we present a distributed version of the resource manager consisting of several ANNs in which each ANN is responsible for modeling application performance and power consumption of a single VM while exchanging information with other ANNs to coordinate resource allocation. Simulated and real experiments show the effectiveness of the distributed ANN resource manager over static allocation, a centralized version and a distributed non-coordinated version.
distributedprocessing in a network of computational cells which realise simple Boolean functions is used as a model for biological processing in neural cell assemblies. Recent work in establishing the criteria for de...
详细信息
distributedprocessing in a network of computational cells which realise simple Boolean functions is used as a model for biological processing in neural cell assemblies. Recent work in establishing the criteria for determining stable behavior is briefly summarized and the implications in characterizing an adaptivity mechanism for such a network are discussed. Empirical results are presented to indicate the deficiencies of typical adaptation algorithms in ensuring stability, and the paper finally points the discussion toward a theory of adaptation which emphasizes topological features of a network rather than individual cellular functions.
Primate vision with its system of interacting subsystems is a prime example of a distributedprocessingneuralnetwork. The available neurophysiological evidence indicates that the functional subsystem concerned with ...
详细信息
Primate vision with its system of interacting subsystems is a prime example of a distributedprocessingneuralnetwork. The available neurophysiological evidence indicates that the functional subsystem concerned with optimal pattern vision sequentially filters the neural image corresponding to retinal input and the filter functions can be described using linear systems analysis.
As recent neuralnetworks are being improved to be more accurate, their model's size is exponentially growing. Thus, a huge number of parameters requires to be loaded and stored from/in memory hierarchy and comput...
详细信息
ISBN:
(纸本)9781728165820
As recent neuralnetworks are being improved to be more accurate, their model's size is exponentially growing. Thus, a huge number of parameters requires to be loaded and stored from/in memory hierarchy and computed in processors to perform training or inference phase of neuralnetworkprocessing. Increasing the number of parameters causes a big challenge for real-time deployment since the memory bandwidth improvement's trend cannot keep up with models' complexity growing trend. Although some operations in neuralnetworks processing are computational intensive such as convolutional layer computing, computing dense layers face with memory bandwidth bottleneck. To address the issue, the paper has proposed Partition Pruning for dense layers to reduce the required parameters while taking into consideration parallelization. We evaluated the performance and energy consumption of parallel inference of partitioned models, which showed a 7.72x speedup of performance and a 2.73x reduction in the energy used for computing pruned fully connected layers in TinyVGG16 model in comparison to running the unpruned model on a single accelerator. Besides, our method showed a limited reduction in accuracy while partitioning fully connected layers.
This paper presents a distributed Denial-of-Services (DDoS) attack detection method using Deep neuralnetwork (DNN) based on flow-based network information. In the proposed method, Radial Basis Function is adopted for...
详细信息
ISBN:
(数字)9786165904773
ISBN:
(纸本)9786165904773
This paper presents a distributed Denial-of-Services (DDoS) attack detection method using Deep neuralnetwork (DNN) based on flow-based network information. In the proposed method, Radial Basis Function is adopted for feature selection to enhance the accuracy of the DNN. The flow-based dataset CICIDS2017 is used for training and testing the detection accuracy. Experimental results show that the DDoS detection accuracy of the proposed method under the dataset are 99.37%.
We propose a Chaotic Associative Memory (CAM) using distributed Patterns for image retrieval. This model is based on the CAM which can separate superimposed patterns and the Multi Winners Self-Organizing neural Networ...
详细信息
ISBN:
(纸本)0780372786
We propose a Chaotic Associative Memory (CAM) using distributed Patterns for image retrieval. This model is based on the CAM which can separate superimposed patterns and the Multi Winners Self-Organizing neuralnetwork which has the ability to generate distributed representation patterns corresponding to input in a self-organizing manner.
暂无评论