The concept of `diversity' has been one of the main open issues in the field of multiple classifier systems. In this paper we address a facet of diversity related to its effectiveness for ensemble construction, namely, explicitly using diversity measures for ensemble construction techniques based on the kind of overproduce and choose strategy known as ensemble pruning. Such a strategy consists of selecting the (hopefully) more accurate subset of classifiers out of an original, larger ensemble. Whereas several existing pruning methods use some combination of individual classifiers' accuracy and diversity, it is still unclear whether such an evaluation function is better than the bare estimate of ensemble accuracy. We empirically investigate this issue by comparing two evaluation functions in the context of ensemble pruning: the estimate of ensemble accuracy, and its linear combination with several well-known diversity measures. This can also be viewed as using diversity as a regularizer, as suggested by some authors. To this aim we use a pruning method based on forward selection, since it allows a direct comparison between different evaluation functions. Experiments on thirty-seven benchmark data sets, four diversity measures and three base classifiers provide evidence that using diversity measures for ensemble pruning can be advantageous over using only ensemble accuracy, and that diversity measures can act as regularizers in this context.
The methane hazard is one of the most dangerous phenomena in hard coal mining. In a certain range of concentrations, methane is flammable and explosive. Therefore, in order to maintain the continuity of the production process and the safety of work for the crew, various measures are taken to prevent these concentration levels from being exceeded. A significant role in this process is played by the forecasting of methane concentrations in mine headings. This very problem has been the focus of the present article. Based on discrete measurements of methane concentration in mine headings and ventilation parameters, the distribution of methane concentration levels in these headings was forecasted. This process was performed on the basis of model-based tests using the Computational Fluid Dynamics (CFD). The methodology adopted was used to develop a structural model of the region under analysis, for which boundary conditions were adopted on the basis of the measurements results in real-world conditions. The analyses conducted helped to specify the distributions of methane concentrations in the region at hand and determine the anticipated future values of these concentrations. The results obtained from model-based tests were compared with the results of the measurements in realworld conditions. The methodology using the CFD and the results of the tests offer extensive possibilities of their application for effective diagnosis and forecasting of the methane hazard in mine headings.
Peat soils (FAO — Gelic Histosols) in the southern Bellsund coast area occur on slopes and terraces. They are formed in places favourable for plant growth, i.e. adequately moistened and fertilized largely with bird excrements. These formations belong to moss peats which are generally decomposed weakly and moderately to about 0.5 m depth. Their content of organic matter is equal to about 30-90%, but it is higher in terrace peats. The latter are more acidified than slope peats. The reaction both of slope and terrace peat soils is as a rule, slightly acid or neutral, and CaC03 content does not exceed 10%. As regards the content of macroelements, that of Al is the highest followed by Ca, Fe, Mg and P. Little K and Ti, and only traces of Na are found. Microelements occur in the following sequence: Mn, Zn, Cu, Cr, Ni, Pb, Co, Cd. Particularly Mn, Zn as well as Cu and Cd were found in a higher concentration. Slope peat soils are richer in macro- and microelements than terrace ones, e.g. 4 times in the case of Mg. Peat soils poor in ash parts (up to 25% ash), contain the fewest elements. Some regularities concern also a vertical distribution of the particular profiles but only with regard to terrace peat soils.
Paper presents the description and floristic and ecologica characteristics of three plant communities on the area of Jasnorzewski Gardens in the region of Arctowski Station (Polish Academy of Sciences) on King George Island. They are: 1) Deschampsio antarctici-Colobanthetum quitensis, 2) Polytrichetum alpini, 3) Calliergidio austro-straminei-Calliergonetum sarmentosi. All communities show a considerable differentiation to several variants. Distribution of plant communities on the studied area is presented on a map based on computer analysis of multispectral air photographs.
Białowieża in Poland is a very famous region in Europe (because of its primeval forest and bison population), but its environmental history is poorly known. This article shows the results of palynological analysis, macrofossil analysis and geological settings of two mires in the Białowieża Forest. The pollen diagrams show changes of the vegetation cover from the younger part of the Late Glacial until the present time. The relative time scale is based on palynostratigraphy and comparison to published results of other sites from the adjacent regions. During the Late Glacial two stages of the vegetation succession were revealed: steppe and forest during the Alleröd period and tundra-like vegetation during the Younger Dryas. The Holocene history consists of five stages of plant cover development. The special features of the Białowieża Forest are conditioned by two main factors: low degree of anthropogenic impact and influences of continental climate and boreal zone, stronger than in the other regions of Poland.
Vegetation changes in the Lower San Valley near Jaroslaw are reconstructed from the Younger Dryas to the present time on the basis of palynological analysis of the peat core. The pollen profile came from a an old riverbed and was supplemented by radiocarbon datings. The Younger Dryas and early Preboreal vegetation was characterised by a high proportion of forest communities with pine (Pinus sylvestris and P. cembra) and birch (Betula), while patches of open area were dominated by the steppe with Artemisia. Climatic amelioration during the Preboreal chronozone led to the rapid spread of elm (Ulmus), which was probably a dominant taxon on the lowest terraces of the valley. Terrestrialization of the water body existing in the palaeomeaner and the subsequent beginning of peat accumulation caused a deterioration in pollen preservation. Hence, the interpretation of the profile section spanning the period between the Boreal and Subatlantic chronozones was seriously disturbed due to selective corrosion and the overrepresentation of Pinus sylvestris type and Filicales monolete sporomorphs. Between ca. 336 and 1152 AD fluctuations in woodland cover were recorded. Important components in those forests, despite the domination of Pinus sylvestris, were Quercus, Carpinus betulus, Fagus sylvatica and Abies alba. The first pollen grains of cereals (Cerealia type) were found before ca. 1605-1414 BC and may be attributed to the agricultural activity of the Neolithic and/or early Bronze tribes. Periods of strong deforestation caused by humans were probably related to the time when the Tarnobrzeska Group of the Lusatian Culture and the Przeworska Culture were active. The first groups of Slavs did not significantly influence the environment, but the subsequent development of those groups led to more visible deforestation, which was triggered after the establishment of Jaroslaw in the 11 th century AD.