-
Article
2D-THA-ADMM: communication efficient distributed ADMM algorithm framework based on two-dimensional torus hierarchical AllReduce
Model synchronization refers to the communication process involved in large-scale distributed machine learning tasks. As the cluster scales up, the synchronization of model parameters becomes a challenging tas...
-
Article
Open AccessA Communication Efficient ADMM-based Distributed Algorithm Using Two-Dimensional Torus Grou** AllReduce
Large-scale distributed training mainly consists of sub-model parallel training and parameter synchronization. With the expansion of training workers, the efficiency of parameter synchronization will be affect...
-
Article
Hybrid MPI/OpenMP parallel asynchronous distributed alternating direction method of multipliers
The distributed alternating direction method of multipliers (ADMM) is one of the most widely used algorithms to solve large-scale optimization problems. Since the memory consumption, communication cost and con...
-
Article
HSAC-ALADMM: an asynchronous lazy ADMM algorithm based on hierarchical sparse allreduce communication
The distributed alternating direction method of multipliers (ADMM) is an effective algorithm for solving large-scale optimization problems. However, its high communication cost limits its scalability. An async...
-
Chapter and Conference Paper
A Dynamic Scheduling Strategy of ADMM Sub-problem Optimization Algorithm Based on Hierarchical Structure
The Alternating Direction Method of Multiplier (ADMM) is a simple algorithm to resolve decomposable convex optimization problems, especially effective in solving large-scale problems. However, this algorithm s...
-
Chapter and Conference Paper
Asynchronous Distributed ADMM for Learning with Large-Scale and High-Dimensional Sparse Data Set
The distributed alternating direction method of multipliers is an effective method to solve large-scale machine learning. At present, most distributed ADMM algorithms need to transfer the entire model paramete...
-
Chapter and Conference Paper
ADMMLIB: A Library of Communication-Efficient AD-ADMM for Distributed Machine Learning
Alternating direction method of multipliers (ADMM) has recently been identified as a compelling approach for solving large-scale machine learning problems in the cluster setting. To reduce the synchronization...
-
Chapter and Conference Paper
Construction of Laboratory Refined Management in Local Applied University
The refined management of laboratory in local applied university is conducive to the formation of mode to innovative talent training and the improvement the ability to innovate. To ensure the efficient operat...
-
Chapter and Conference Paper
Collaborative Computing of Urban Built-Up Area Identification from Remote Sensing Image
Urban built-up area is one of the important criterions of urbanization. Remote sensing can quickly acquire dynamic temporal and spatial variation of urban built-up area, but how to identify and extract urban b...
-
Article
Open AccessResearch on digital image encryption algorithm based on double logistic chaotic map
With the development of information technology, image information has become the main content of network information transmission. With the development of image encryption technology, it is also about the deve...
-
Chapter and Conference Paper
Fast Communication Structure for Asynchronous Distributed ADMM Under Unbalance Process Arrival Pattern
The alternating direction method of multipliers (ADMM) is an algorithm for solving large-scale data optimization problems in machine learning. In order to reduce the communication delay in a distributed enviro...
-
Article
Volcanic ash cloud detection from MODIS image based on CPIWS method
Volcanic ash cloud detection has been a difficult problem in moderate-resolution imaging spectroradiometer (MODIS) multispectral remote sensing application. Principal component analysis (PCA) and independent c...
-
Chapter
Improving Content Recommendation in Social Streams via Interest Model
The current microblog recommendation approaches mainly consider users’ interests. But because user’s interests are changing dynamically and they have low activity, it’s hard to build user interest model. In th...
-
Article
Open AccessBSP-Based Support Vector Regression Machine Parallel Framework
In this paper, we investigate the distributed parallel Support Vector Machine training strategy, and then propose a BSP-Based Support Vector Regression Machine Parallel Framework which can implement the most o...
-
Chapter and Conference Paper
A Novel Parallel Interval Exclusion Algorithm
The optimization algorithm based on interval analysis is a deterministic global optimization algorithm. However, Solving high-dimensional problems, traditional interval algorithm exposed lots of problems such ...
-
Chapter and Conference Paper
Scalable SPMD Algorithm of Evolutionary Computation
This paper addresses two parallelization techniques used for evolutionary computation. We study the grid enabled evolutionary computation model, and the differences between the coevolution and the space decomp...
-
Chapter and Conference Paper
The System for Computing of Molecule Structure on the Computational Grid Environment
The computing grid can offer users tremendous computer resources. In this paper, we proposed an application model for computing of molecule structure on the computational grid. According to the need of grid co...
