Comparing Human and Neural Network Learning of Climate Categories     

Robert Lloyd  Gregory Carbone 《The Professional geographer》1995,47(3):237-250

This study compares how humans and neural networks classify climate types. Human subjects were asked to classify climates from monthly temperature and precipitation patterns. To model their learning process, the same data were used to produce input vectors that trained a pattern associator neural network. Both human subjects and the neural network classified climates accurately after 10 rounds of supervised learning. The neural network successfully modeled the rate of human learning and the ability to learn specific climate categories. Moreover, the neural network weights used to classify climates correspond to distinct visual characteristics in temperature and precipitation. These results suggest that neural networks can model the formation of visual categories.  相似文献   

水库群优化调度函数的人工神经网络方法研究   总被引：17，自引：0，他引：17       下载免费PDF全文

胡铁松  万永华 《水科学进展》1995,6(1):53-60

提出了研究水库群优化调度函数的人工神经网络方法,并探讨了神经网络的训练参数、训练方法和训练样本的改变对网络训练和应用效果的影响。实例研究表明,模型及其算法是可行的、有效的。  相似文献   

Assessment of reliability in water distribution networks using entropy based measures   总被引：2，自引：0，他引：2  

K. Awumah  I. Goulter  S. K. Bhatt 《Stochastic Environmental Research and Risk Assessment (SERRA)》1990,4(4):309-320

Entropy based expressions for measurement of reliability and redundancy have recently been reported. These measures approach assessment of the reliability of the distribution network from the intrinsic redundancy of the network layout. The paper extends earlier work on entropy functions by including a more explicit statement of the alternate paths available in the network and by recognizing that under certain circumstances, e.g., failure of some part of the network work, an outflow link from a node under normal working condition may become an inflow link to the same node. The measures are assessed by comparison with parameters measuring Nodal Pair Reliability and percentage of flow supplied at adequate pressure for a range of networks and link failure conditions in this networks. The entropy measures are shown to reflect changes in the network reliability, as measured by these two comparative parameters, very well.  相似文献   

Cost-effective network design for groundwater flow monitoring   总被引：1，自引：0，他引：1  

R. Andricevic 《Stochastic Environmental Research and Risk Assessment (SERRA)》1990,4(1):27-41

The extensive use of groundwater resources has increased the need for developing cost-effective monitoring networks to provide an indication of the degree to which the subsurface environment has been affected by human activities. This study presents a cost-effective approach to the design of groundwater flow monitoring networks. The groundwater network design is formulated with two problem formats: maximizing the statistical monitoring power for specified budget constraint and minimizing monitoring cost for statistical power requirement. The statistical monitoring power constraint is introduced with an information reliability threshold value. A branch and bound technique is employed to select the optimal solution from a discrete set of possible network alternatives. The method is tested to the design of groundwater flow monitoring problem in the Pomona County, California.  相似文献   

Results from a new model of river basin evolution   总被引：1，自引：0，他引：1  

Garry Willgoose  Rafael L. Bras  Ignacio Rodriguez-Iturbe 《地球表面变化过程与地形》1991,16(3):237-254

This paper briefly describes a model of the erosional development of catchments and their channel networks. The model differentiates between the dominant transport processes in hillslope and channels. The development of channels and hillslopes occurs in an integrated manner as a function of physically observable mechanisms. The growth of a river basin is qualitatively described. The model concepts are used to study the basin during periods of growth (transient periods), as well as during dynamic equilibrium. This leads to hypotheses about the relationship between slopes, relief, tectonic uplift, erodability, runoff, and catchment area. It is shown that the model leads to very reasonable and desirable behaviour of hillslopes during retreat and degradation.  相似文献   

Applying　artificial　neural　networks　to　achieve　quantitative　division　of　potential　focal　regions     

姜效典  张维岗 《地震学报(英文版)》1994,7(4):645-652

（姜效典，张维岗）Ａｐｐｌｙｉｎｇａｒｔｉｆｉｃｉａｌｎｅｕｒａｌｎｅｔｗｏｒｋｓｔｏａｃｈｉｅｖｅｑｕａｎｔｉｔａｔｉｖｅｄｉｖｉｓｉｏｎｏｆｐｏｔｅｎｔｉａｌｆｏｃａｌｒｅｇｉｏｎｓ￥Ｘｉａｏ－ＤｉａｎＪＩＡＮＧａｎｄＷｅｉ－ＧａｎｇＺＨＡＮＧ（...  相似文献   

人工神经网络在天气预报中的应用研究   总被引：14，自引：4，他引：14  

张承福 《气象》1994,20(6):43-47

介绍了将人工神经元网络用于灾害性天气（暴雨）预报研究的情况，分析了天气预报问题的数学提法及困难所在，神经元网络用于天气预报的原理，暴雨预报的特点及我们对网络模型的改进。结果表明，神经风格确可通过学习从原始数据中提取足够的分类信息，达到较好的预报准确率，值得进一步研究。  相似文献   

神经网络方法在地震预报研究中的初步应用     

冯德益  汪德馨 《地震》1994,(4):23-29

本文把神经网络方法引进地震预报研究当中。使用地震频次，最大震级，平均震级，等价地震次数等多项地震活动性指标作为神经网络的输入，未来时段内的最大地震震级作为其输出，可以对某一固定地区的最大地震震级作出中近期预报。选用的神经网络模型为含两个中间层的前向模型，并采用ＢＰ算法。所得结果表明，用神经网络方法可以在一定精度范围内使震级预报的内检符合率达到１００％，在本文的例子中，外推预报准确率达到６０％以上。  相似文献   

BP网络在位场解释中的应用     

朱晓军  申宁华 《吉林大学学报(地球科学版)》1995,(1)

介绍了ＢＰ网络的特点，阐述了它在位场资料解释中应用的方法、特点和可行性。通过模型计算证实了该方法在位场异常识别中的效果，正确率达９７％。此外，通过修改网络单元的激发函数，增强网络的稳定性和内插功能，使ＢＰ网络在多层密度界面反演中得到满意的结果，模型和实例计算证明了该方法的有效性。  相似文献   

膨胀土判别与分类的人工神经网络方法   总被引：18，自引：0，他引：18  

易顺民  晏同珍 《地质科技情报》1995,14(3):91-94

提出了运用神经网络理论对膨胀土进行判别与分类的方法。通过对安康膨胀土的实测数据分析，建立了研究膨胀土判别与分类的计算机智能专家系统。结果表明，用人工神经网络方法对膨胀土进行判别与分类的准确率很高，它无需在判别因子与预测目标之间建立基于某种理论的经验统计关系，该方法具有重要的实用价值。  相似文献