Five hundred and eight phytosociological relevés from pine forests on sand, calcareous gravel and rock in NE and S Germany were analysed with respect to the frequency of Ellenberg indicator values of vascular plants for nutrients (N). Principal component analysis revealed that after the average nitrogen value (mN), the distribution shape and modality are the second most important sources of variation in the N-spectra of relevés. Of the five spectral types defined by combinations of mN and modality, the unimodal low nutrient type (66.5%) prevailed, followed by bimodal distributions with many indicators for low and high N-supply, with few in the intermediate classes 4 and 5 (27.4%), whereas spectra with a single mode at high (3.9%) or intermediate (2.2%) N- values were rare. Two explanations for the frequent coexistence of vascular plant indicators of N-deficiency with those indicating eutrophication are discussed: (a) Bimodality may be a consequence of the low capacity of pine forests to sequestre the excess input of anthropogenic nitrogen from the atmosphere, and/or (b) the natural dynamics of humus accumulation and mineralization following disturbance. To avoid misinterpretation of mN, inspection of modality of the N-spectra should be standard practice when analysing pine forest or other long-lived vegetation with low N-sequestration. Predominance of high N- over low N-indicators in relevés may be interpreted as a signal of advanced anthropogenic eutrophication, N-saturation and increased risk of N-leaching to groundwater. Bimodal spectra with prevailing deficiency indicators, on the other hand, may be either due to short-term N-release or indicate the beginning of eutrophication.
The unsupervised learning of feature extraction in high-dimesional patterns is a central problem for the neural network approach. Feature extraction is a procedure which maps original patterns into the feature (or factor) space of reduced dimension. In this paper we demonstrate that Hebbian learning in Hopfield-like neural network is a natural procedure for unsupervised learning of feature extraction. Due to this learning, factors become the attractors of network dynamics, hence they can be revealed by the random search. The neurodynamics is analysed by Single-Step approximation which is known [8] to be rather accurate for sparsely encoded Hopfield-network. Thus, the analysis is restricted by the case of sparsely encoded factors. The accuracy of Single-Step approximation is confirmed by Computer simulations.
We deal with a sequencing problem that arises when there are multiple repair actions available to fix a broken man-made system and the true cause of the system failure is uncertain. The system is formally described by a probabilistic model, and it is to be repaired by a sequence of troubleshooting actions designed to identify the cause of the malfunction and fix the system. The task is to find a course of repair with minimal expected cost. We propose a binary integer programming formulation for the problem. This can be used to solve the problem directly or to compute lower bounds of the minimal expected cost using linear programming relaxation. We also present three greedy algorithms for computing initial feasible solutions.
Purpose of this work is to show that the Particle Swarm Optimization Algorithm may improve the results of some well known Machine Learning methods in the resolution of discrete classification problems. A binary version of the PSO algorithm is used to obtain a set of logic rules that map binary masks (that represent the attribute values), to the available classes. This algorithm has been tested both in a single pass mode and in an iterated mode on a well-known set of problems, called the MONKS set, to compare the PSO results against the results reported for that domain by the application of some common Machine Learning algorithms.
We introduce the function Z(x;ξ,ν):=∫x−∞φ(t−ξ)⋅Φ(νt)dt, where φ and Φ are the pdf and cdf of N(0,1), respectively. We derive two recurrence formulas for the effective computation of its values. We show that with an algorithm for this function, we can efficiently compute the second-order terms of Bonferroni-type inequalities yielding the upper and lower bounds for the distribution of a max-type binary segmentation statistic in the case of small samples (where asymptotic results do not work), and in general for max-type random variables of a certain type. We show three applications of the method -- (a) calculation of critical values of the segmentation statistic, (b) evaluation of its efficiency and (c) evaluation of an estimator of a point of change in the mean of time series.
The influence of monopolar binaural galvanic stimulation of the vestibular system was studied on body sway. Subjects, with eyes closed, were standing on a hard support or on foam rubber. Their body sway was registered on a force platform at intervals of 50 s. Both polarities of direct current with intensity 1 mA were used as a galvanic stimulus during the whole recording interval. Changes of body sway amplitude and velocity were analyzed in situations with and without galvanic stimulation on two different support surfaces. In stance on the hard support, the cathodal polarization of labyrinths (in most subjects) reduced body sway velocity and decreased body sway slightly in the anteroposterior direction. Anodal polarization of labyrinths during 50 s did not affect the body sway parameters. The results on the foam rubber platform exhibited a significant reduction of body sway velocity induced by both anodal and cathodal polarization of the labyrinths. The decrease of body sway in the anteroposterior direction was also observed during cathodal polarization. The stabilizing effect of vestibular binaural monopolar stimulation on the upright stance was mainly observed in the postural control situation where the leg proprioceptive input was changed (stance on soft surface) and the role of vestibular input was more important.
Phenylcyclohexylglycoloylester of hydroxyethyldimethylhydrazonium (compound VGFB-3113) has been shown earlier to have a strong antimuscarinic effect on smooth muscle. Its affinity to muscarinic binding sites in homogenates of rat heart ventricles (M2 subtype), submandibular salivary gland (M3 subtype) and brain cortex (predominantly Ml subtype) has now been investigated in radioligand displacement experiments using (3H)quinuclidinyl benzilate ((3H)QNB) as a relatively non-specific muscarinic ligand. VGFB-3113 inhibited the binding of (3H)QNB with pKj values of 8.17, 8.73, and 8.52 in the heart, salivary gland, and brain cortex, respectively. It is concluded that the compound has a high affinity for muscarinic binding sites without strong preference for any of the Ml-M3 subtypes.