Probabilistic learning on manifolds (PLoM) with partition

The probabilistic learning on manifolds (PLoM) introduced in 2016 has solved difficult supervised problems for the “small data” limit where the number N of points in the training set is small. Many extensions have since been proposed, making it possible to deal with increasingly complex cases. However, the performance limit has been observed and explained for applications for which N is very small and for which the dimension of the diffusion-map basis is close to N. For these cases, we propose a novel extension based on the introduction of a partition in independent random vectors. We take advantage of this development to present improvements of the PLoM such as a simplified algorithm for constructing the diffusion-map basis and a new mathematical result for quantifying the concentration of the probability measure in terms of a probability upper bound. The analysis of the efficiency of this extension is presented through two applications.     

Some Grubbs-Type Statistics for the Detection of Several Outliers

Several widely used tests for outlying observations are reviewed. Problems of repeated application and “masking” are described. Suggested as appropriate to over-come these problems are two new statistics: L_k which is based on the k largest (observed) values and E_k which is based on the k largest (in absolute value) residuals. Tables of approximate critical values for these statistics are given for 0.01, .025,0.05, and 0.10 levels of significance and for sample size n = 3 (1) 20 (5) 50.     

Inductive transfer for learning Bayesian networks

In several domains it is common to have data from different, but closely related problems. For instance, in manufacturing, many products follow the same industrial process but with different conditions; or in industrial diagnosis, where there is equipment with similar specifications. In these cases it is common to have plenty of data for some scenarios but very little for others. In order to learn accurate models for rare cases, it is desirable to use data and knowledge from similar cases; a technique known as transfer learning. In this paper we propose an inductive transfer learning method for Bayesian networks, that considers both structure and parameter learning. For structure learning we use conditional independence tests, by combining measures from the target task with those obtained from one or more auxiliary tasks, using a novel weighted sum of the conditional independence measures. For parameter learning, we propose two variants of the linear pool for probability aggregation, combining the probability estimates from the target task with those from the auxiliary tasks. To validate our approach, we used three Bayesian networks models that are commonly used for evaluating learning techniques, and generated variants of each model by changing the structure as well as the parameters. We then learned one of the variants with a small dataset and combined it with information from the other variants. The experimental results show a significant improvement in terms of structure and parameters when we transfer knowledge from similar tasks. We also evaluated the method with real-world data from a manufacturing process considering several products, obtaining an improvement in terms of log-likelihood between the data and the model when we do transfer learning from related products.     

Performance comparison of the CRAY X-MP/24 with SDD and the CRAY-2

The CRAY-2 is considered to be one of the most powerful supercomputers. Its state-of-the-art technology features a faster clock and more memory than any other supercomputer available today. In this report the single processor performance of the CRAY-2 is compared with the older, more mature CRAY X-MP. Benchmark results are included for both the slow and the fast memory DRAM MOS CRAY-2. Our comparison is based on a kernel benchmark set aimed at evaluating the performance of these two machines on some standard tasks in scientific computing. Particular emphasis is placed on evaluating the impact of the availability of large real memory on the CRAY-2 versus fast secondary memory on the CRAY X-MP with SSD. Our benchmark includes large linear equation solvers and FFT routines, which test the capabilities of the different approaches to providing large memory. We find that in spite of its higher processor speed the CRAY-2 does not perform as well as the CRAY X-MP on the Fortran kernel benchmark. We also find that for large-scale applications, which have regular and predictable memory access patterns, a high-speed secondary memory device such as the SSD can provide performance equal to the large real memory of the CRAY-2.The author is an employee of SCA Division of Boeing Computer Services.     

Towards van der Waals Epitaxial Growth of GaAs on Si using a Graphene Buffer Layer

Van der Waals growth of GaAs on silicon using a two‐dimensional layered material, graphene, as a lattice mismatch/thermal expansion coefficient mismatch relieving buffer layer is presented. Two‐dimensional growth of GaAs thin films on graphene is a potential route towards heteroepitaxial integration of GaAs on silicon in the developing field of silicon photonics. Hetero‐layered GaAs is deposited by molecular beam epitaxy on graphene/silicon at growth temperatures ranging from 350 °C to 600 °C under a constant arsenic flux. Samples are characterized by plan‐view scanning electron microscopy, atomic force microscopy, Raman microscopy, and X‐ray diffraction. The low energy of the graphene surface and the GaAs/graphene interface is overcome through an optimized growth technique to obtain an atomically smooth low­ temperature GaAs nucleation layer. However, the low adsorption and migration energies of gallium and arsenic atoms on graphene result in cluster‐growth mode during crystallization of GaAs films at an elevated temperature. In this paper, we present the first example of an ultrasmooth morphology for GaAs films with a strong (111) oriented fiber‐texture on graphene/silicon using quasi van der Waals epitaxy, making it a remarkable step towards an eventual demonstration of the epitaxial growth of GaAs by this approach for heterogeneous integration.     

Characterizing executive functioning in older special populations: From cognitively elite to cognitively impaired.

The authors examined the structure and invariance of executive functions (EF) across (a) a continuum of cognitive status in 3 groups of older adults (cognitively elite [CE], cognitively normal [CN], and cognitively impaired [CI]) and (b) a 3-year longitudinal interval. Using latent variable analyses (LISREL 8.80), the authors tested 3-factor models (“Inhibition”: Hayling [Burgess & Shallice, 1997], Stroop [Regard, 1981]; “Shifting”: Brixton [Burgess & Shallice, 1997], Color Trails [D’Elia et al., 1996]; and “Updating”: Reading and Computational Span [Salthouse & Babcock, 1991]) and 1-factor models within each group. Participants (initial N = 570; 53–90 years) were from the Victoria Longitudinal Study (Sample 3, Waves 1 and 2). Cross-sectionally, the authors observed a 3-factor EF structure especially for the CE group and 1-factor solutions for all 3 groups. Longitudinally, temporal invariance was supported for the 3-factor model (CE and CN groups) and the 1-factor model (CI and CN groups). Subgroups with higher cognitive status and greater 3-year stability performed better on EF factors than corresponding groups with lower cognitive status and less stability. Studies of EF structure, performance, dedifferentiation, and dysfunction will benefit from considering initial cognitive status and longitudinal stability. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Copper accumulation in soils from two different‐aged apricot orchards in Central Otago,New Zealand

In recent years there has been a move in New Zealand to convert small, uneconomic orchards to other agricultural uses. Overseas research has demonstrated that copper from fungicide sprays often accumulates in the soils of orchards and can cause certain problems for plant growth and livestock. To date, no assessments appear to have been made of the degree of copper accumulation in such older orchards in New Zealand. This study presents the results of an assessment of two well‐established apricot orchards in the Central Otago region of South Island, New Zealand.

Two aspects of the results are surprising. The first is the relatively low levels of copper in the soil when compared with results from another orchard in the region. That orchard is 18 years old and mean soil copper has already reached 100.0 μg/g, indicating a much faster rate of accumulation than in the two orchards in the current study. The second unexpected result is that the younger orchard has a significantly higher level of copper than the older orchard. Reasons for these findings are discussed in relation to apparent differences in soil properties.     

Strategies in environmental education

The causes of environmental changes, including climate, are neither fully understood nor identified, nor do scientists agree about their nature. Yet, their impact is such that action cannot be delayed. This involves spreading public awareness through environmental education and sensitization. Hence a need to reach, to educate the poorly informed public at large. With the classroom as the base, “town meetings” and eco‐tourism may constitute potentially useful approaches towards that goal.     

Water treatment with silicofluorides and lead toxicity

Toxic metals like lead, manganese, copper and cadmium damage neurons and deregulate neurotransmitters like serotonin and dopamine (which are essential to normal impulse control and learning). Earlier studies show that — controlling for socio‐economic and demographic factors — environmental pollution with lead is a highly significant risk factor in predicting higher rates of crime, attention deficit disorder or hyperactivity, and learning disabilities. Exposure and uptake of lead has been associated with industrial pollution, leaded paint and plumbing systems in old housing, lead residues in soil, dietary habits (such as shortages of calcium and iron), and demographic factors (such as poverty, stress, and minority ethnicity). We report here on an additional “risk co‐factor” making lead and other toxic metals in the environment more dangerous to local residents: the use of silicofluorides as agents in water treatment. The two chemicals in question — fluosilicic acid and sodium silicofluoride — are toxins that, despite claims to the contrary, do not dissociate completely and change water chemistry when used under normal water treatment practices. As a result, water treatment with siliconfluorides apparently functions to increase the cellular uptake of lead. Data from lead screening of over 280,000 children in Massachusetts indicates that silicofluoride usage is associated with significant increases in average lead in children's blood as well as percentage of children with blood lead in excess of 10μg/dL. Consistent with the hypothesized role of silicofluorides as enhancing uptake of lead whatever the source of exposure, children are especially at risk for higher blood lead in those communities with more old housing or lead in excess of 15 ppb in first draw water samples where silicofluorides are also in use. Preliminary findings from county‐level data in Georgia confirm that silicofluoride usage is associated with higher levels of lead in children's blood. In both Massachusetts and Georgia, moreover, behaviors associated with lead nurotoxicity are more frequent in communities using silicofluorides than in comparable localities that do not use these chemicals. Because there has been insufficient animal or human testing of silicofluoride treated water, further study of the effect of silicofluorides is needed to clarify the extent to which these chemicals are risk co‐factors for lead uptake and the hazardous effects it produces.     

How Brittany and Florida coasts cope with green tides

Although the direct involvement of nitrogen and phosphorus has been shown, eutrophication remains poorly managed to this day. The excessive growth of some opportunist seaweeds is the consequence in coastal ecosystems close to agricultural or strongly urbanized and industrialized zones. In Brittany, a leading tourist region of France, green tides set down on the beaches big quantities of Ulva, of which some 100,000 m³ are harvested annually, with significant ecological and economic consequences. In Florida, although the macroalgae (notably Codium) drift about with the currents, they are of sufficient mass and spread over such areal extent to inhibit penetration of sunlight through the water column to the coral reef surface. They strand also on the beaches. Stabilization of algae by composting and methanization of hydrolyzed and pressed algae juice are two methods which have been studied in Brittany and can be used to enhance the value of the harvested seaweed, depending on local conditions and on evolution of needs in energy, basic materials or organic products. But the parallel made here between two very different ecosystems, both disturbed, leads above all to the question of what course to follow now to avoid a worldwide disaster.

Bien que le rôle de l’azote et du phosphore dans les processus d’eutrophication ait été montré, celle‐ci reste mal contrôlée à ce jour. La croissance excessive d’algues opportunistes en est la conséquence dans des écosystèmes côtiers proches de zones agricoles ou fortement urbanisées et industrialisées. En Bretagne, région touristique de France, des marées vertes déversent sur les plages de grandes quantités d’ulves, dont environ 100,000 m³ sont récoltés annuellement, avec d’importantes conséquences écologiques et économiques. En Floride, bien que les macroalgues (notamment Codium) dérivent avec les courants, elles sont en masse suffisante et suffisamment réparties pour empêcher la pénétration de la lumière dans la colonne d’eau jusqu’au récif de corail. Elles s’échouent également sur les plages. La stabilisation par compostage et la méthanisation de jus d’algues hydrolysées et pressées sont deux méthodes qui ont été étudiées en Bretagne afin de valoriser l’algue récoltée, en fonction des conditions locales et de l’évolution des besoins en énergie, matières premières ou produits organiques. Mais le parallèle fait ici entre deux écosystèmes très différents, tous deux perturbés, mène surtout à se demander quelle voie suivre maintenant pour éviter un désastre mondial.