Mixtures of probabilistic principal component analyzers model high-dimensional nonlinear data by combining local linear models. Each mixture component is specifically designed to extract the local principal orientations in the data. An important issue with this generative model is its sensitivity to data lying off the low-dimensional manifold. In order to address this problem, the mixtures of robust probabilistic principal component analyzers are introduced. They take care of atypical points by means of a long tail distribution, the Student-t. It is shown that the resulting mixture model is an extension of the mixture of Gaussians, suitable for both robust clustering and dimensionality reduction. Finally, we briefly discuss how to construct a robust version of the closely related mixture of factor analyzers. 相似文献
Creep experiments were conducted on ice crystals in compression to investigate the effects of boundary conditions on a single-slip system deformed in plane strain. Friction at the platens of the deformation apparatus introduces a bending moment which causes a variation in the amount of lattice rotation across the specimen. This is shown to occur in mechanically constrained crystals observed through plane polarized light. Relieving the constraints and minimizing friction at the ice-platen contact leads to the widening of the sample near the specimen-platen interface and the production of tails symmetrically disposed about the longitudinal axis of the deformed crystals. This is interpreted to originate from a bending moment in the opposite sense from that obtained in the constrained crystals, resulting from a progressive increase in slip displacement towards the platens where the segments of the slip plane become shorter. When the crystal ends were constrained but allowed to move sideways, a simple shear regime was established in which lattice slip was concentrated in the centre of the crystal. 相似文献
Journal of Intelligent Manufacturing - Remanufacturing includes disassembly and reassembly of used products to save natural resources and reduce emissions. While assembly is widely understood in... 相似文献
The classification task usually works with flat and batch learners, assuming problems as stationary and without relations between class labels. Nevertheless, several real-world problems do not assume these premises, i.e., data have labels organized hierarchically and are made available in streaming fashion, meaning that their behavior can drift over time. Existing studies on hierarchical classification do not consider data streams as input of their process, and thus, data is assumed as stationary and handled through batch learners. The same can be said about works on streaming data, as the hierarchical classification is overlooked. Studies concerning each area individually are promising, yet, do not tackle their intersection. This study analyzes the main characteristics of the state-of-the-art works on hierarchical classification for streaming data concerning five aspects: (i) problems tackled, (ii) datasets, (iii) algorithms, (iv) evaluation metrics, and (v) research gaps in the area. We performed a systematic literature review of primary studies and retrieved 3,722 papers, of which 42 were identified as relevant and used to answer the aforementioned research questions. We found that the problems handled by hierarchical classification of data streams include mainly classification of images, human activities, texts, and audio; the datasets are mostly created or synthetic data; the algorithms and evaluation metrics are well-known techniques or based on those; and research gaps are related to dynamic context, data complexity, and computational resources constraints. We also provide implications for future research and experiments to consider common characteristics shared amongst hierarchical classification and data stream classification.
We investigated whether, in rheumatoid arthritis (RA), the CD45 isoform expression of peripheral blood T-lymphocytes (T-PBL) is related to auto-immune processes (e.g. IgM rheumatoid factors) and to clinical manifestations. By three-colour flow cytometry, we quantified three subsets of CD4+ or CD8+ T-PBL: "naive" CD45RA+,RO-, "transient" CD45RA+,RO+, and "memory" CD45RA-,RO+ cells, in 102 patients with RA and in 41 age- and sex-matched controls. The serum levels of rheumatoid factors (RF) were determined--besides conventional agglutination tests--by ELISA (IgM-RF). Extensive clinical examination was performed at the time of blood sampling. In RA, age, sex and drug therapy did not constitute major influences on the CD45RA/RO patterns. In "healthy" men, higher age significantly' correlated with fewer naive and more memory CD4+ T-PBL (P < 0.01). In RA, distinct correlations between the T-PBL subsets, autoimmune and clinical manifestations became obvious when patients with low and high levels of RF against human IgG Fc fragments, as determined by ELISA, were analysed separately. RA patients with high IgM-RF had elevated proportions of CD45RO+ T-PBL (P < 0.05), that correlated with clinical parameters of disease activity (tender joint count, Ritchie index, P < 0.05) and outcome (Health Assessment Questionnaire, Larsen radiographic scores, P < 0.05). The proportions of memory CD4+ and CD8+ T-PBL correlated strongly (P < 0.001) with the IgM-RF levels. Within 1 year, only three of 34 patients (disease duration of 5-9 years) showed seroconversion from low to high levels of IgM-RF (and positive agglutination tests); this was paralleled by reductions in naive and increases in transient T-PBL (P < 0.02). Thus, in RA, the proportions of memory CD4+ and CD8+ T-PBL correlate with the level of IgM-RF and, together with transient T-PBL, with clinical parameters of disease activity and outcome. 相似文献
Exploring the power of shared memory communication objects and models, and the limits of distributed computability are among the most exciting research areas of distributed computing. In that spirit, this paper focuses on a problem that has received considerable interest since its introduction in 1987, namely the renaming problem. It was the first non-trivial problem known to be solvable in an asynchronous distributed system despite process failures. Many algorithms for renaming and variants of renaming have been proposed, and sophisticated lower bounds have been proved, that have been a source of new ideas of general interest to distributed computing. It has consequently acquired a paradigm status in distributed fault-tolerant computing.In the renaming problem, processes start with unique initial names taken from a large name space, then deciding new names such that no two processes decide the same new name and the new names are from a name space that is as small as possible.This paper presents an introduction to the renaming problem in shared memory systems, for non-expert readers. It describes both algorithms and lower bounds. Also, it discusses strong connections relating renaming and other important distributed problems such as set agreement and symmetry breaking. 相似文献
We generalize the Kleene theorem to the case where nonassociative products are used. For this purpose, we apply rotations restricted to the root of binary trees. 相似文献
Appropriate information on solar resources is very important for a variety of technological areas, such as: agriculture, meteorology, forestry engineering, water resources and in particular in the designing and sizing of solar energy systems. However, the availability of observed solar radiation measurements has proven to be spatially and temporally inadequate for many applications. In this paper we propose to merge the global solar radiation measurements from the Royal Meteorological Institute of Belgium solar measurements network with the operationally derived surface incoming global short-wave radiation products from Meteosat Second Generation satellites imageries to improve the spatio-temporal resolution of the surface global solar radiation data over Belgium. We evaluate several merging methods with various degrees of complexity (from mean field bias correction to geostatistical merging techniques) together with interpolated ground measurements and satellite-derived values only. The performance of the different methods is assessed by leave-one-out cross-validation. 相似文献
We present in this paper a model for indexing and querying web pages, based on the hierarchical decomposition of pages into blocks. Splitting up a page into blocks has several advantages in terms of page design, indexing and querying such as (i) blocks of a page most similar to a query may be returned instead of the page as a whole (ii) the importance of a block can be taken into account, as well as (iii) the permeability of the blocks to neighbor blocks: a block b is said to be permeable to a block b?? in the same page if b?? content (text, image, etc.) can be (partially) inherited by b upon indexing. An engine implementing this model is described including: the transformation of web pages into blocks hierarchies, the definition of a dedicated language to express indexing rules and the storage of indexed blocks into an XML repository. The model is assessed on a dataset of electronic news, and a dataset drawn from web pages of the ImagEval campaign where it improves by 16% the mean average precision of the baseline. 相似文献