An online self-organizing scheme for Parsimonious and Accurate Fuzzy Neural Networks

In this paper, an online self-organizing scheme for Parsimonious and Accurate Fuzzy Neural Networks (PAFNN), and a novel structure learning algorithm incorporating a pruning strategy into novel growth criteria are presented. The proposed growing procedure without pruning not only simplifies the online learning process but also facilitates the formation of a more parsimonious fuzzy neural network. By virtue of optimal parameter identification, high performance and accuracy can be obtained. The learning phase of the PAFNN involves two stages, namely structure learning and parameter learning. In structure learning, the PAFNN starts with no hidden neurons and parsimoniously generates new hidden units according to the proposed growth criteria as learning proceeds. In parameter learning, parameters in premises and consequents of fuzzy rules, regardless of whether they are newly created or already in existence, are updated by the extended Kalman filter (EKF) method and the linear least squares (LLS) algorithm, respectively. This parameter adjustment paradigm enables optimization of parameters in each learning epoch so that high performance can be achieved. The effectiveness and superiority of the PAFNN paradigm are demonstrated by comparing the proposed method with state-of-the-art methods. Simulation results on various benchmark problems in the areas of function approximation, nonlinear dynamic system identification and chaotic time-series prediction demonstrate that the proposed PAFNN algorithm can achieve more parsimonious network structure, higher approximation accuracy and better generalization simultaneously.     

对金龟子具有杀虫专一性的苏云金芽孢杆菌综述(英文)

自从 190 1年Ishiwata首次报导猝倒菌以来 ,已有数千篇有关苏云金芽孢杆菌的论文相继发表。在苏云金芽孢杆菌 (Bt)的害虫生物防治领域中首次发现对鳞翅目昆虫有杀虫专一性的Bt菌株以后 ,又有了划时代的进展。Goldberg和Margalit发现了杀双翅目幼虫的以色列亚种 (subsp .isrealensis) ,以及Kreig等人发现了对某些鞘翅目幼虫有毒的拟步行甲亚种 (sbusp .tenebrionis)。 1992年 ,Ohba等分离到了日本亚种(subsp .japonensis)对金龟子有专一性杀虫作用的菌株Buibui。该菌株及与之有关的研究进展即是本文讨论的重点。这些发现不仅丰富了苏云金芽孢杆菌科学 ,也丰富了生物防治的应用策略。另一方面 ,在毒素蛋白结构及其杀虫机制的分子生物学和生物化学领域也已获巨大进展。但对杀虫机制尚未完全明了。例如 ,对杀虫蛋白与中肠上皮细胞原生质膜上的受体蛋白的相互作用就知之甚少。这一领域在某种程度上还是众说纷纭。展论这些令人兴奋而又难解的课题将最终促成二十一世纪可持续农业的建立。     

Screening of the insecticidal activity of Bacillus thuringiensis strains against Lygus hesperus (Hemiptera: Miridae) nymphal population

Lygus hesperus Knight (Hemiptera: Miridae) is an economically important insect pest controlled primarily by chemical pesticides. Bacillus thuringiensis Berliner is a gram-positive bacterium that has been developed for the control of some insect pests in the orders Lepidoptera, Coleoptera, and Diptera. In this study, whole culture extracts of 94 B. thuringiensis strains from 83 serovars were added to an artificial diet and assayed against L. hesperus first and second instars. A total of five B. thuringiensis strains, B. thuringiensis variety thuringiensis, thuringiensis exotoxin +, morrisoni, tolworthi, and darmstadiensis generated > 98% mortality after 7 d of incubation. The screening was repeated with 117 alkali-solubilized trypsin-digested B. thuringiensis cultures and the same five B. thuringiensis strains showed nearly identical results. All five strains produce beta-exotoxin, which exhibits a wide host spectrum activity. No beta-exotoxin-minus B. thuringiensis strains showed significant toxicity against L. hesperus nymphs. The present work is one of the first thorough screenings of the wide diversity of the B. thuringiensis varieties for the control of L. hesperus nymphal populations.     

Nonlinear system modelling via optimal design of neural trees

This paper introduces a flexible neural tree model. The model is computed as a flexible multi-layer feed-forward neural network. A hybrid learning/evolutionary approach to automatically optimize the neural tree model is also proposed. The approach includes a modified probabilistic incremental program evolution algorithm (MPIPE) to evolve and determine a optimal structure of the neural tree and a parameter learning algorithm to optimize the free parameters embedded in the neural tree. The performance and effectiveness of the proposed method are evaluated using function approximation, time series prediction and system identification problems and compared with the related methods.     

Generalization in Adaptation to Stable and Unstable Dynamics

Humans skillfully manipulate objects and tools despite the inherent instability. In order to succeed at these tasks, the sensorimotor control system must build an internal representation of both the force and mechanical impedance. As it is not practical to either learn or store motor commands for every possible future action, the sensorimotor control system generalizes a control strategy for a range of movements based on learning performed over a set of movements. Here, we introduce a computational model for this learning and generalization, which specifies how to learn feedforward muscle activity in a function of the state space. Specifically, by incorporating co-activation as a function of error into the feedback command, we are able to derive an algorithm from a gradient descent minimization of motion error and effort, subject to maintaining a stability margin. This algorithm can be used to learn to coordinate any of a variety of motor primitives such as force fields, muscle synergies, physical models or artificial neural networks. This model for human learning and generalization is able to adapt to both stable and unstable dynamics, and provides a controller for generating efficient adaptive motor behavior in robots. Simulation results exhibit predictions consistent with all experiments on learning of novel dynamics requiring adaptation of force and impedance, and enable us to re-examine some of the previous interpretations of experiments on generalization.     

Preparation of spray-dried wettable powder formulations of Bacillus thuringiensis-based biopesticides

Bacillus thuringiensis is the most widely used biopesticide among many methods available to control insects. To make a saleable product, B. thuringiensis must be substantially concentrated by removal of water and formulated to improve longevity, efficacy, and ease of transport of the product. B. thuringiensis subsp. aizawai culture broth as an active ingredient was mixed with various adjuvants and then spray dried. The optimum conditions for spray drying were found to be an outlet temperature of 60-85 degrees C and an inlet temperature of 120-180 degrees C. Various adjuvants had different effects on physical and biological properties of the dried product. Gelatinized tapioca starch and milk powder improved suspensibility but adversely affected wettability of the dried formulated product. Vegetable oil and Tween 20 enhanced wettability but resulted in poor suspensibility. Silica fume was used to enhance flowability because it reduced clumping and caking of the powder resulting from the addition of vegetable oil. Formulation containing 10% wt:wt B. thuringiensis, 10% wt:wt gelatinized tapioca starch, 10% wt:wt sucrose, 38% wt:wt tapioca starch, 20% wt:wt milk powder, 10% wt:wt silica fume, 2% wt:wt polyvinyl alcohol, 5% vol:vol Tween 20, 1% vol:vol refined rice bran oil, and 1% vol:vol antifoam solution was found to be optimum in terms of the physical and biological properties of the dried product. This formulation had 55% suspensibility, 24 s for wetting time, and 5.69 x 10(4) CFU/ml of LC50 value against Spodoptera exigua larvae.     

基于径向基函数神经网络的温室室内温度预测模型

试验证实径向基函数神经网络(Radial Basias Function Neural Network)在函数逼近能力、训练速度方面都有良好的性能．采用最小正交二乘法为训练算法,基于传统的数学分析,用PRIVA公司温室监控系统采集数据,选用当前时刻室外温度、风速、太阳辐照度、顶窗开度、内帘幕展开度、水温、室内温度、相对湿度,再加上1个时间间隔、2个时间间隔以前的室内温度作为输入向量,获得了满意的温室室内温度一步预测模型(均方差等于0．0073)．该模型为设计温室环境控制器及分析温室性能奠定了良好基础．     

A novel efficient learning algorithm for self-generating fuzzy neural network with applications

In this paper, a novel efficient learning algorithm towards self-generating fuzzy neural network (SGFNN) is proposed based on ellipsoidal basis function (EBF) and is functionally equivalent to a Takagi-Sugeno-Kang (TSK) fuzzy system. The proposed algorithm is simple and efficient and is able to generate a fuzzy neural network with high accuracy and compact structure. The structure learning algorithm of the proposed SGFNN combines criteria of fuzzy-rule generation with a pruning technology. The Kalman filter (KF) algorithm is used to adjust the consequent parameters of the SGFNN. The SGFNN is employed in a wide range of applications ranging from function approximation and nonlinear system identification to chaotic time-series prediction problem and real-world fuel consumption prediction problem. Simulation results and comparative studies with other algorithms demonstrate that a more compact architecture with high performance can be obtained by the proposed algorithm. In particular, this paper presents an adaptive modeling and control scheme for drug delivery system based on the proposed SGFNN. Simulation study demonstrates the ability of the proposed approach for estimating the drug's effect and regulating blood pressure at a prescribed level.     

Generalized Verlet Algorithm for Efficient Molecular Dynamics Simulations with Long-range Interactions

For the purpose of molecular dynamics simulations of large biopolymers we have developed a new method to accelerate the calculation of long-range pair interactions (e.g. Coulomb interaction). The algorithm introduces distance classes to schedule updates of non-bonding interactions and to avoid unnecessary computations of interactions between particles which are far apart. To minimize the error caused by the updating schedule, the Verlet integration scheme has been modified. The results of the method are compared to those of other approximation schemes as well as to results obtained by numerical integration without approximation. For simulation of a protein with 12 637 atoms our approximation scheme yields a reduction of computer time by a factor of seven. The approximation suggested can be implemented on sequential as well as on parallel computers. We describe an implementation on a (Transputer-based) MIMD machine with a systolic ring architecture.     

Biological control of Dutch elm disease: Bacillus thuringiensis as a potential control agent for Scolytus scolytus and S. multistriatus

J assim , H.K., F oster , H.A. & F airhurst , C.P. 1990. Biological control of Dutch elm disease: Bacillus thuringiensis as a potential control agent for Scolytus scolytus and S. multistriatus. Journal of Applied Bacteriology 69 , 563–568.
The effects of exposing fifth instar larvae of Scolytus scolytus and S. multistriatus to spore suspensions of Bacillus spp. were investigated. Bacillus thuringiensis ser 3a, 3b increased the mortality of larvae cultured on an artificial medium from approximately 20% in control cultures to over 80% in cultures exposed to the bacteria. The mortality was dose-dependent for S. multistriatus and the approximate LC₅₀ value was 2.2 times 10³ spores/ml. Different serotypes of B. thuringiensis caused different levels of mortality: H6 produced the highest mortality and H1 the lowest. Bacillus alvei and B. cereus were also pathogenic but B. megaterium was not. The results are discussed in relation to the mechanism of pathogenicity and the potential for the use of B. thuringiensis for the control of the vectors of Dutch elm disease.     

黑龙江省苏云金杆菌的分离及对夜蛾科昆虫高毒力菌株的筛选

【目的】为了发掘新的苏云金杆菌(Bacillus thuringiensis)的资源,在黑龙江省不同地区采集不同类型的土壤样品分离出对夜蛾科具有高毒力的菌株。【方法】采用醋酸钠选择性筛选法筛选Bt菌株,利用10对通用引物对分离株进行基因型分析,SDS-PAGE进行杀虫晶体蛋白分析,同时测定苏云金杆菌分离株对棉铃虫Helicoverpa armigera(Hübner)、甘蓝夜蛾Mamestra brassicae(Linnaeus)、斜纹夜蛾Spodoptera litura(Fabricius)的杀虫活性。【结果】从黑龙江省不同地区采集的352份不同类型的土壤样品中,共分离出46株苏云金芽孢杆菌野生菌株,出菌率为13.06%。油镜下可观察到伴孢晶体的形态有菱形、球形、镶嵌形及不规则形。结果表明产菱形晶体的菌株多含有cry1类基因,而同时产生菱形、球形及不规则形晶体的菌株则含有多种基因型。SDS-PAGE蛋白分析发现这些菌株主要表达130、90、60 ku蛋白。对其中的部分菌株进行毒力测定,结果表明有4株菌株对3种夜蛾科昆虫具有高毒力。【结论】黑龙江省苏云金芽孢杆菌分布广泛,类型多样,已获得对夜蛾科昆虫有高毒力的菌株,这对夜蛾科害虫的绿色防控及延缓其抗性具有重要意义。     

Specificity and efficacy of purified Bacillus thuringiensis proteins against agronomically important insects

The host range and relative efficacy of three purified Bacillus thuringiensis insect control proteins were determined against 17 different agronomically important insects representing five orders and one species of mite. The three B. thuringiensis proteins were single gene products from B. thuringiensis ssp. kurstaki HD-1 (CryIA(b)) and HD-73 (CryIA(c)), both lepidopteran-specific proteins, and B. thuringiensis ssp. tenebrionis (CryIIIA), a coleopteran-specific protein. Seven insects showed sensitivity to both B. thuringiensis ssp. kurstaki proteins, whereas only 1 of the 18 insects was sensitive to B. thuringiensis ssp. tenebrionis protein. The level of B. thuringiensis ssp. kurstaki protein required for 50% mortality (LC50) varied by 2000-fold for these 7 insects. A larval growth inhibition assay was developed to determine the amount of B. thuringiensis ssp. kurstaki protein required to inhibit larval growth by 50% (EC50). This extremely sensitive assay enabled detection of B. thuringiensis ssp. kurstaki HD-73 levels as low as 1 ng/ml.     

Efficient expression of vip184DeltaP gene under the control of promoters plus Shine-Dalgarno (SD) sequences of cry genes from Bacillus thuringiensis

AIMS: To compare vip184DeltaP gene expression time course and Vip184 protein yield under the control of promoters and Shine-Dalgarno (SD) sequences of vip184, cry3A and cry1A gene from Bacillus thuringiensis respectively. METHODS AND RESULTS: Derived from the shuttle vector pHT3101, recombinant plasmids pHPT3, pHTP3A(Delta)P and pHTP1A(Delta)P were constructed with the native vip184 gene and the vip184(Delta)P gene, either under the control of promoters and SD sequences of cry3A or cry1A genes. When the above plasmids were transformed into an acrystalliferous B. thuringiensis strain Cry(-)B, their expression time course were consistent with those of vip184, cry3A and cry1A gene respectively. The maximum yields of Vip184 protein were increased when under the control of promoters plus SD sequences of cry3A and cry1A gene. CONCLUSIONS: The results showed that both cry3A and cry1A promoter/SD sequence combinations were able to enhance synthesis of Vip184 and change its expression time course. SIGNIFICANCE AND IMPACT OF THE STUDY: Both cry3A and cry1A promoter/SD systems offer a method for improving the expression efficacy of the vip184 gene in B. thuringiensis and it is possible to co-express the vip184 gene and cry genes and accumulate Vip184 in the form of inclusion bodies by these systems in order to construct novel useful B. thuringiensis engineered strains.     

Generating balanced learning and test sets for function approximation problems

In function approximation problems, one of the most common ways to evaluate a learning algorithm consists in partitioning the original data set (input/output data) into two sets: learning, used for building models, and test, applied for genuine out-of-sample evaluation. When the partition into learning and test sets does not take into account the variability and geometry of the original data, it might lead to non-balanced and unrepresentative learning and test sets and, thus, to wrong conclusions in the accuracy of the learning algorithm. How the partitioning is made is therefore a key issue and becomes more important when the data set is small due to the need of reducing the pessimistic effects caused by the removal of instances from the original data set. Thus, in this work, we propose a deterministic data mining approach for a distribution of a data set (input/output data) into two representative and balanced sets of roughly equal size taking the variability of the data set into consideration with the purpose of allowing both a fair evaluation of learning's accuracy and to make reproducible machine learning experiments usually based on random distributions. The sets are generated using a combination of a clustering procedure, especially suited for function approximation problems, and a distribution algorithm which distributes the data set into two sets within each cluster based on a nearest-neighbor approach. In the experiments section, the performance of the proposed methodology is reported in a variety of situations through an ANOVA-based statistical study of the results.     

Distribution of Bacillus thuringiensis subsp. israelensis in Soil of a Swiss Wetland reserve after 22 years of mosquito control

Recurrent treatments with Bacillus thuringiensis subsp. israelensis are required to control the floodwater mosquito Aedes vexans that breeds in large numbers in the wetlands of the Bolle di Magadino Reserve in Canton Ticino, Switzerland. Interventions have been carried out since 1988. In the present study, the spatial distribution of resting B. thuringiensis subsp. israelensis spores in the soil was measured. The B. thuringiensis subsp. israelensis concentration was determined in soil samples collected along six transects covering different elevations within the periodically flooded zones. A total of 258 samples were processed and analyzed by quantitative PCR that targeted an identical fragment of 159 bp for the B. thuringiensis subsp. israelensis cry4Aa and cry4Ba genes. B. thuringiensis subsp. israelensis spores were found to persist in soils of the wetland reserve at concentrations of up to 6.8 log per gram of soil. Continuous accumulation due to regular treatments could be excluded, as the decrease in spores amounted to 95.8% (95% confidence interval, 93.9 to 97.7%). The distribution of spores was correlated to the number of B. thuringiensis subsp. israelensis treatments, the elevation of the sampling point, and the duration of the flooding periods. The number of B. thuringiensis subsp. israelensis treatments was the major factor influencing the distribution of spores in the different topographic zones (P < 0.0001). These findings indicated that B. thuringiensis subsp. israelensis spores are rather immobile after their introduction into the environment.     

Learning, memory, and the role of neural network architecture

The performance of information processing systems, from artificial neural networks to natural neuronal ensembles, depends heavily on the underlying system architecture. In this study, we compare the performance of parallel and layered network architectures during sequential tasks that require both acquisition and retention of information, thereby identifying tradeoffs between learning and memory processes. During the task of supervised, sequential function approximation, networks produce and adapt representations of external information. Performance is evaluated by statistically analyzing the error in these representations while varying the initial network state, the structure of the external information, and the time given to learn the information. We link performance to complexity in network architecture by characterizing local error landscape curvature. We find that variations in error landscape structure give rise to tradeoffs in performance; these include the ability of the network to maximize accuracy versus minimize inaccuracy and produce specific versus generalizable representations of information. Parallel networks generate smooth error landscapes with deep, narrow minima, enabling them to find highly specific representations given sufficient time. While accurate, however, these representations are difficult to generalize. In contrast, layered networks generate rough error landscapes with a variety of local minima, allowing them to quickly find coarse representations. Although less accurate, these representations are easily adaptable. The presence of measurable performance tradeoffs in both layered and parallel networks has implications for understanding the behavior of a wide variety of natural and artificial learning systems.     

An efficient algorithm for approximating geodesic distances in tree space

The increasing use of phylogeny in biological studies is limited by the need to make available more efficient tools for computing distances between trees. The geodesic tree distance-introduced by Billera, Holmes, and Vogtmann-combines both the tree topology and edge lengths into a single metric. Despite the conceptual simplicity of the geodesic tree distance, algorithms to compute it don't scale well to large, real-world phylogenetic trees composed of hundred or even thousand leaves. In this paper, we propose the geodesic distance as an effective tool for exploring the likelihood profile in the space of phylogenetic trees, and we give a cubic time algorithm, GeoHeuristic, in order to compute an approximation of the distance. We compare it with the GTP algorithm, which calculates the exact distance, and the cone path length, which is another approximation, showing that GeoHeuristic achieves a quite good trade-off between accuracy (relative error always lower than 0.0001) and efficiency. We also prove the equivalence among GeoHeuristic, cone path, and Robinson-Foulds distances when assuming branch lengths equal to unity and we show empirically that, under this restriction, these distances are almost always equal to the actual geodesic.     

Computational modeling of evolutionary learning processes in the brain

The evolutionary selection circuits model of learning has been specified algorithmically. The basic structural components of the selection circuits model are enzymatic neurons, that is, neurons whose firing behavior is controlled by membrane-bound macromolecules called excitases. Learning involves changes in the excitase contents of neurons through a process of variation and selection. In this paper we report on the behavior of a basic version of the learning algorithm which has been developed through extensive interactive experiments with the model. This algorithm is effective in that it enables single neurons or networks of neurons to learn simple pattern classification tasks in a number of time steps which appears experimentally to be a linear function of problem size, as measured by the number of patterns of presynaptic input. The experimental behavior of the algorithm establishes that evolutionary mechanisms of learning are competent to serve as major mechanisms of neuronal adaptation. As an example, we show how the evolutionary learning algorithm can contribute to adaptive motor control processes in which the learning system develops the ability to reach a target in the presence of randomly imposed disturbances.     

Evaluation of oils and microbial pathogens for control of lepidopteran pests of sweet corn in New England

Vegetable and mineral oil, Beauveria bassiana (Balsamo) and Bacillus thuringiensis subsp. kurstaki Berliner were evaluated for control of Ostrinia nubilalis (Hübner) (Lepidoptera: Pyralidae), Helicoverpa zea (Boddie) (Lepidoptera: Noctuidae) and Spodoptera frugiperda (J. E. Smith) (Lepidoptera: Noctuidae) in sweet corn (Zea mays L.). Field experiments in Maine and Massachusetts during 1993 and 1994 evaluated oils and pathogens singly or in combinations, using a single application directly to the top of the silk channel, immediately after pollination. Mineral oil alone provided equal (1993) or better (1994) control compared with corn oil. In both years, mineral or corn oil plus B. thuringiensis resulted in 93-98% marketable ears, compared with 48-52% marketable ears in untreated plots. In three factorial experiments with B. bassiana, B. thuringiensis and corn oil, B. bassiana at 5 x 10(7) conidia per ear provided little or no control while B. thuringiensis and corn oil provided significant though not always consistent control of all three species. The combination of B. thuringiensis and corn oil provided the largest and most consistent reduction in numbers of larvae and feeding damage to ears.     

绞股蓝对电休克大鼠记忆障碍的改善作用

绞股蓝又名七叶胆,为葫芦科绞股蓝属植物。它具有多种药理活性,并有实验证明其具有增强学习记忆的功能。本实验主要验证绞股蓝对电休克法造成的SD大鼠记忆障碍的改善作用。实验方法分为四个过程:①建立大鼠的空间分辨学习记忆;②采用电休克的方法破坏大鼠的记忆,建立记忆遗忘的大鼠模型;③把记忆遗忘的大鼠模型随机分成两组,分别用2.5%绞股蓝提取物的溶液和0.9%生理盐水(每天每只8ml/100g)对两组模型大鼠进行灌胃饲养8天;④观察测试大鼠的空间分辨学习记忆的改善情况,用Y型迷宫模型测试出记忆错误次数、正确率、测试总时间和主动回避次数等参考指标。结果表明绞股蓝组与对照组相比,错误次数较少,正确率较高,测试总时间较短,主动回避次数较多。由此证明绞股蓝对SD大鼠记忆障碍有明显的改善作用。