SYMBIOSIS OF HETEROGENEOUS PARALLELISMS

异构并行性的共生

• 批准号: 04650301
• 负责人: MATSUYAMA Yasuo
• 金额: $ 1.41万
• 依托单位: IBARAKI UNIVERSITY
• 依托单位国家: 日本
• 项目类别: Grant-in-Aid for General Scientific Research (C)
• 财政年份: 1992
• 资助国家: 日本
• 起止时间: 1992 至 1993
• 项目状态: 已结题
• 来源: https://kaken.nii.ac.jp/en/grant/KAKENHI-PROJECT-04650301/
• 关键词: Heterogeneous parallelism; Connectionism; Massive parallelism; Learning; Multiple criteria optimization; log-bias; Mutation; Harmonic competition; 教師あり学習; 教師なし学習; ダイバージェンス; 二重並列性; 相利共生; 競合学習; ニューラルネット; 計算論的学習; 自己組織化

This study has a dual purpose : Designing an emulator which realizes symbiosis of heterogeneous parallelisms and presenting new connectionst learning algorithms. On the realization of the emulator, two workstations are used. One is for an SIMD mechanism where a finegrained parallelism is emulated. The other is for a coarse-grained parallelsm which controls the massive parallel part. KL1 was used for this control mechanism. The multiply descent cost competitive learning algorithm was run on this symbiotic system. The nondeterminism caused by the parallelsm was found to be rather meritorious for the exit from bad local minima.For the developement of new learning algorithms, the head investigator presented two major new methods. On the supervised learning, the backpropagation with additional penalties was presented. This algorithm includes entropy/divergence penalties on the weithts and outputs. Pruning of the network and improvement of errors and generalization were acheived.On the unsupervised case, the head investigator created the harmonic competitive learning. This algorithm enables to solve multiple criteria optimization with the aid of self-organization. The logarithmic competition bias and the logarithmic weight mutation solved the local optimality in the case of data compression.Thus, this research project was completed by accomplishing the claimed results.

本研究具有双重目的：设计一个实现异构并行共生的仿真器，并提出新的连接学习算法。在仿真器的实现上，使用了两个工作站。一个是SIMD机制，其中模拟了细粒度的并行性。另一种是控制大量并行部分的粗粒度并行。KL1用于这种控制机制。在该共生系统上运行了多重下降代价竞争学习算法。发现由并行性引起的不确定性对于从坏的局部极小值中退出是很有价值的。对于新的学习算法的发展，首席研究员提出了两种主要的新方法。在监督学习上，提出了带附加惩罚的反向传播算法。该算法在权重和输出上包含熵/散度惩罚。实现了网络的修剪、误差的改善和泛化。在无监督情况下，首席调查员创造了和谐竞争学习。该算法能够借助自组织解决多准则优化问题。对数竞争偏差和对数权值突变解决了数据压缩时的局部最优性问题。因此，这个研究项目完成了所声称的结果。

项目成果

Y.Matsuyama: "Competitive Learning among Massively Parallel Agents" Neural, Parallel & Scientific Computations. vol.1. 181-198 (1993)

Y.Matsuyama：“大规模并行智能体之间的竞争学习”神经、并行

Y.Matsuyama: "Learning in Competitive Networks with Penalties" Proc.Int.Joint Conf.on Neural Networks. IV. 773-778 (1992)

Y.Matsuyama：“在带有惩罚的竞争网络中学习”Proc.Int.Joint Conf.on Neural Networks。

Yasuo Matsuyama: "Learning in competitive networks with penalties" Proc.Int.Joint.Conf.on Neural Networks. IV. 773-778 (1992)

Yasuo Matsuyama：“在有惩罚的竞争网络中学习”Proc.Int.Joint.Conf.on Neural Networks。

Y.Matsuyama: "Competitive Learning among Massively Parallel Agents" Neural,Parallel & Scientific Computations. I. 181-198 (1993)

Y.Matsuyama：“大规模并行代理之间的竞争学习”神经，并行

Y.Matsuyama & M.Tan: "Digital Movies Using Optimized Feature Maps" Proc.Int.Conference on Neural Networks. x. x-y (1994)

松山勇