SOIL MOISTURE PREDICTION USING SUPPORT VECTOR MACHINES1

ABSTRACT: Herein, a recently developed methodology, Support Vector Machines (SVMs), is presented and applied to the challenge of soil moisture prediction. Support Vector Machines are derived from statistical learning theory and can be used to predict a quantity forward in time based on training that uses past data, hence providing a statistically sound approach to solving inverse problems. The principal strength of SVMs lies in the fact that they employ Structural Risk Minimization (SRM) instead of Empirical Risk Minimization (ERM). The SVMs formulate a quadratic optimization problem that ensures a global optimum, which makes them superior to traditional learning algorithms such as Artificial Neural Networks (ANNs). The resulting model is sparse and not characterized by the “curse of dimensionality.” Soil moisture distribution and variation is helpful in predicting and understanding various hydrologic processes, including weather changes, energy and moisture fluxes, drought, irrigation scheduling, and rainfall/runoff generation. Soil moisture and meteorological data are used to generate SVM predictions for four and seven days ahead. Predictions show good agreement with actual soil moisture measurements. Results from the SVM modeling are compared with predictions obtained from ANN models and show that SVM models performed better for soil moisture forecasting than ANN models.     

在数学教学中如何培养学生能力

学生能力的培养是教学活动的中心,本文通过对数学教学过程的总结和分析,就如何培养学生能力问题谈了自己的一些见解,并针对学习能力和理解能力的培养提出了一些措施。     

模糊神经网络用于环境大气质量评价的研究

研究了基于多准则学习的模糊神经网络评价环境大气质量的模型,并以此对某矿区的环境大气质量进行了评价,结果表明该方法简便、实用、客观。     

Web数据挖掘技术在远程教育中的应用

使用Web数据挖掘技术发挥网络教育中资源共享以及无时间局限性的优势,达到帮助学习者快速有效地寻找满足其个性需求的教学资源的目的,并且对其学习进程进行准确指导.     

Resource distributions and diet development by trial-and-error learning

We study interactions between resource distributions, grouping, and diet development in foragers who learn by trial-and-error. We do this by constructing an individual-based model where individuals move and forage in groups in a 2-D space with high resource diversity and learn what to eat. By comparing diet development in different resource distributions, and in gregarious and solitary individuals, we elucidate how these factors affect patterns of diet variation. Our results indicate that different resource distributions have profound effects on learning opportunities, and thereby lead to contrasting phenomena. In uniform environments, local resource depletion by gregarious individuals, in interaction with learning, leads to diet differentiation. In patchy environments, grouping leads to enhanced diet overlap within groups and leads to differences in diet between groups. Surprisingly, mixed environments can generate all these phenomena simultaneously. Our results predict relationships between diet variation, trial-and-error learning, and resource distributions. The phenomena we describe are not evolved strategies, but arise spontaneously when groups of individuals learn to forage in certain resource distributions. This suggests that describing diet specialization or diet homogenization as the result of behavioral strategies may not always be justified.     

Learning to predict channel stability using biogeomorphic features

Current human land use activities are altering many components of the river landscape, resulting in unstable channels. Instability may have serious negative consequences for water quality, aquatic and riparian habitat, and for river-related human infrastructure such as bridges and roads. Resource management agencies have developed rapid bioassessment surveys to help assess stability in a fast and cost-effective way. While this assessment can be done for a single site fairly rapidly, it is still time-consuming to apply over large watersheds and assessment activities must be prioritized. We constructed a system that employs commonly available map data as inputs to cost-sensitive variants of decision tree algorithms to predict the relative channel stability of different sites. In particular, we use bagged lazy option trees (LOTs) and bagged probability estimation trees (PETs) to identify all unstable channels while making the smallest number of errors of classifying stable channels as unstable, thereby minimizing cost and maximizing safety. We measured the performance of the classifiers using ROC curves and found that the PETs performed better than the LOTs in situations where the number of instances of the stable and unstable classes were relatively balanced, but the LOTs did better where unstable examples were relatively rare compared to stable, perhaps due to the LOTs’ ability to focus on individual examples.     

Less is more: social learning of predator recognition requires a low demonstrator to observer ratio in Arctic charr (Salvelinus alpinus)

Reintroduction programs for conservation purposes suffer from low post-release survival of captive-bred animals, especially so with fish reintroductions. High mortality is most often due to weakened antipredator skills, which could be enhanced through direct experience with predators and social learning. As yet, the relative efficiency of these learning methods has not been tested, and the significance of the factors determining how efficiently antipredator responses are transmitted socially from experienced (demonstrator) to predator naive (observer) individuals is largely unknown. Using hatchery-bred stock of the endangered Saimaa Arctic charr (Salvelinus alpinus), we show for the first time that: (1) contrary to theoretical predictions, acquired predator recognition was only socially transmitted from predator-experienced to predator-naive fish in groups with a smaller proportion of experienced than naive individuals, and that (2) socially acquired avoidance of the predator odors was as strong as in those fish conditioned with a combination of odors and predator exposure. We conclude that using social transmission procedures in training fish for reintroduction programs may provide considerable ecological, economical and ethical advantages.Communicated by J. Krause     

The dynamics of social learning in an insect model,the bumblebee (<Emphasis Type="Italic">Bombus terrestris</Emphasis>)

Bumblebees (Bombus terrestris) are attracted to those particular inflorescences where other bees are already foraging, a process known as local enhancement. Here, we use a quantitative analysis of learning in a foraging task to illustrate that this attraction can lead bees to learn more quickly which flower species are rewarding if they forage in the company of experienced conspecifics. This effect can also be elicited by model bees, rather than live demonstrators. We also show that local enhancement in bumblebees most likely reflects a general attraction to conspecifics that is not limited to a foraging context. Based on the widespread occurrence of both local enhancement and associative learning in the invertebrates, we suggest that social influences on learning in this group may be more common than the current literature would suggest and that invertebrates may provide a useful model for understanding how learning processes based on social information evolve.     

Does colour matter? The importance of colour in avoidance learning, memorability and generalisation

Aposematic species exploit the ability of predators to associate, for example, conspicuous colouration with the unprofitability of prey. We tested the importance of colour for avoidance learning, memory and generalisation in wild-caught great tits (Parus major). First, we determined the birds’ initial colour preferences for red, yellow, orange and grey artificial prey items. The birds showed some preferences, as they were more willing to eat grey prey as their first choice, but these were not strong preferences. We then trained birds to discriminate red, yellow or variable (red and yellow) signals from grey where colours signalled palatable and unpalatable food. In general, the birds learned the discrimination task equally well, irrespective of which colours signalled unpalatability, and subsequently remembered the distinction between previously palatable and previously unpalatable colours in the memorability test. We did not find strong evidence that variability in the signal affected learning or memory. Our results suggest that, in a task where birds must discriminate between palatable and unpalatable prey, it does not matter which specific colour signals unpalatability, although this might be context-dependent. To study whether training also affects responses to unconditioned stimuli, we included orange prey items in the memorability test. Although orange had been palatable in the initial preference test, the birds ate fewer orange prey items after they had been trained to avoid red, yellow or both colours (variable signal) as unpalatable prey, but did not change their preference when trained that these colourful signals were palatable. This indicates that generalisation occurred more readily after a negative experience than a positive experience, a situation that would potentially allow imperfect mimicry to occur.     

Homopteran chemical signatures reduce aggression of tending ants

Summary. Because generalist ants are aggressive towards foreign insects, the recognition of homopterans by tending ants is critical in ant/homopteran trophobiosis. Herein we report experimental evidence indicating that Argentine ant, Linepithema humile (Mayr) (Hymenoptera: Formicidae) learn to associate the production of honeydew with the chemical characteristics of homopteran cuticle, suppressing ant aggression and allowing the ants to tend homopterans. Although chemically-mediated associative learning is well understood in honeybee foraging, to our knowledge, it has not been reported before in ant/homopteran trophobiosis.