The accuracy of the neural networks can usually be improved by increasing the size of the dataset and the layers or operators of the network, as it has strong composability. But, it makes a challenge to train these mo...
详细信息
The rapid advancement of generative artificial intelligence (GAI) has led to the creation of transformative applications such as ChatGPT, which significantly boosts text processing efficiency and diversifies audio, im...
详细信息
The inherent computational complexity of validating and verifying concurrent systems implies a need to be able to exploit parallel and distributedcomputing architectures. We present a new distributed algorithm for st...
详细信息
ISBN:
(纸本)9781665401623
The inherent computational complexity of validating and verifying concurrent systems implies a need to be able to exploit parallel and distributedcomputing architectures. We present a new distributed algorithm for state space exploration of concurrent systems on computing clusters. Our algorithm relies on Remote Direct Memory Access (RDMA) for low-latency transfer of states between computing elements, and on state reconstruction trees for compact representation of states on the computing elements themselves. For the distribution of states between computing elements, we propose a concept of state stealing. We have implemented our proposed algorithm using the OpenSHMEM API for RDMA and experimentally evaluated it on the Grid'5000 testbed with a set of benchmark models. The experimental results show that our algorithm scales well with the number of available computing elements, and that our state stealing mechanism generally provides a balanced workload distribution.
One of the main hurdles of Industry 4.0 is to predict Aircraft failures Since downtime costs money and reduces productivity, the ability to prevent failures is essential. Knowing how many cycles or RULs remain before ...
详细信息
In this paper, we propose a distributed reservoir-computing based parallel nonlinear equalization for 100 Gb/s vertical cavity surface emitting laser (VCSEL) enabled optical interconnects. Equalization performance of ...
详细信息
ISBN:
(纸本)9781665481557
In this paper, we propose a distributed reservoir-computing based parallel nonlinear equalization for 100 Gb/s vertical cavity surface emitting laser (VCSEL) enabled optical interconnects. Equalization performance of proposed equalizer is compared with neural network and Volterra series based equalizers and similar performance can be achieved but with very neat and low computational complexity training process. Moreover, this approach, explained as small reservoirs make a mickle, is a scalable network generation solution that is promising for parallel hardware implementation.
The vehicular adhoc networks (VANETs) is an integral part of the intelligent transportation system required for delivering safety and non-safety messages to vehicles in real-time via vehicle to vehicle and vehicle to ...
详细信息
The DRM (Digital Rights Management) systems protect owners' copyrights by controlling consumers' access to digital works. However, they fail to provide authorization evidence if customers use digital works on ...
详细信息
The proceedings contain 46 papers. The topics discussed include: undervolting on wireless sensor nodes: a critical perspective;a scalable causal broadcast that tolerates dynamics of mobile networks;robotic sorting on ...
ISBN:
(纸本)9781450395601
The proceedings contain 46 papers. The topics discussed include: undervolting on wireless sensor nodes: a critical perspective;a scalable causal broadcast that tolerates dynamics of mobile networks;robotic sorting on the grid;greedy algorithms for scheduling package delivery with multiple drones;a study on migration scheduling in distributed stream processing engines;distributed matrix tiling using a hypergraph labeling formulation;a lattice linear predicate parallel algorithm for the dynamic programming problems;a fast wait-free multi-producers single-consumer queue;limited associativity makes concurrent software caches a breeze;and topology inference of networks utilizing rooted spanning tree embeddings.
Deep learning is a vital technology in our lives today. Both the size of training datasets and neural networks are growing to tackle more challenging problems with deep learning. distributed deep neural network (DDNN)...
详细信息
Predicting resource usage of workloads in large scale production clusters is very important to understand the characteristics of applications. It is also very important for cluster operators to manage cluster resource...
详细信息
ISBN:
(纸本)9781665473156
Predicting resource usage of workloads in large scale production clusters is very important to understand the characteristics of applications. It is also very important for cluster operators to manage cluster resources more efficiently. Traditional statistical-based prediction methods face challenges in predicting resource usage in large scale dynamic and complex clusters. The current commonly used deep learning methods such as Recurrent Neural networks (RNN) usually use the historical data in single node to predict the future resource usage. While most of the modern applications (e.g., microservices) are distributively deployed to the cluster, the traditional single node resource prediction methods cannot predict the resource usages well. To solve this problem, in this paper we propose a new deep learning model called GraphGRU which is based on graph attention networks (GAT) to predict the resource usages from the cluster perspective. We use the Dynamic Time Warping (DTW) algorithm to construct a graph structure for multiple physical nodes in the cluster and also use a method similar to data compensation to co-train the model with both horizontal and vertical data. We validate our model on the Alibaba microservices dataset which is captured from a large scale production cluster. Compared to the traditional deep learning methods, our model improves the prediction accuracy by up to 48.27%.
暂无评论