An empirical investigation of bias and variance in time series forecasting: modeling considerations and error evaluation

Bias and variance play an important role in understanding the fundamental issue of learning and generalization in neural network modeling. Several studies on bias and variance effects have been published in classification and regression related research of neural networks. However, little research has been done in this area for time-series modeling and forecasting. We consider modeling issues related to understanding error components given the common practices associated with neural-network time-series forecasting. We point out the key difference between classification and time-series problems in consideration of the bias-plus-variance decomposition. A Monte Carlo study on the role of bias and variance in neural networks time-series forecasting is conducted. We find that both input lag structure and hidden nodes are important in contributing to the overall forecasting performance. The results also suggest that overspecification of input nodes in neural network modeling does not impact the model bias, but has significant effect on the model variance. Methods such as neural ensembles that focus on reducing the model variance, therefore, can be valuable and effective in time-series forecasting modeling.     

Interaction of the σ2 Receptor Ligand PB28 with the Human Nucleosome: Computational and Experimental Probes of Interaction with the H2A/H2B Dimer

Sigma‐2 (σ₂) binding sites are an emerging target for anti‐neoplastic agents due to the strong apoptotic effect exhibited by σ₂ agonists in vitro and the overexpression of these sites in tumor cells. Nonetheless, no σ₂ receptor protein has been identified. Affinity chromatography using the high‐affinity σ₂ ligand PB28 and human SK‐N‐SH neuroblastoma cells was previously utilized to identify σ₂ ligand binding proteins, specifically histones H1, H2A, H2B, and H3.3a. To rationalize this finding, homology modeling and automated docking studies were employed to probe intermolecular interactions between PB28 and human nucleosomal proteins. These studies predicted interaction of PB28 with the H2A/H2B dimer at a series of sites previously found to be implicated in chromatin compaction and nucleosomal assembly. To experimentally verify this prediction, a competitive binding assay was performed on the reconstituted H2A/H2B dimer using [³H]PB28 as radioligand, and an IC₅₀ value of 0.50 nM was determined for PB28 binding. In addition, [³H]PB28 was found to accumulate with up to a fivefold excess in nuclear fractions over cytosolic fractions of SK‐N‐SH and MCF7 cells, indicating that PB28 is capable of entering the nucleus to interact with histone proteins. In conjunction with computational results, these data suggest that PB28 may exert its cytotoxic effect through direct interaction with nuclear material.     

An Investigation of Group Leaders' Intentions.

Stimulated recall methods were used to assist 34 group leaders in recalling their intentions (i.e., purposes, plans, goals) in relation to interventions they made in a group session. A total of 835 intentions were reported by the leaders and were categorized into 6 separate intention categories by trained judges. The categories included directing the group, gathering information and assessing members, challenging members, attending to and validating members' experiences, directing self, and promoting connections and interactions among group members. Multidimensional scaling procedures revealed 4 intention clusters: Promoting Insight/Change, Planning and Guiding, Attending, and Assessing Growth. Limitations of the study are addressed, along with implications for future research and group leader training. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Fire Performance Assessment of a Fiber Reinforced Polymer Wall Panel Used in a Single Family Dwelling

This paper describes a fire performance assessment of a sandwich panel selected as the primary material in a high-performance house developed for the 2013 China Solar Decathlon competition. The sandwich panel has fiber reinforced polymer skins and a foam plastic core. Evaluation of fire performance showed that the selected panel would not comply with U.S. building code requirements for a single-family dwelling without additional protection. However, the use of an intumescent fire protective coating, coupled with the installation of a residential fire sprinkler system, is shown to be effective in meeting the intent of the fire and life safety requirements of the International Residential Code. Fire testing shows that the application of an intumescent fire protective coating to the panels significantly decreases the heat release rate, the smoke production characteristics, and the extent of flame spread on the panels.     

Determination of the TE and TM modes in arbitrarily shaped waveguides using a hypersingular boundary element formulation

In this paper a procedure based on the hypersingular element method is applied to find the TE and TM modes in arbitrarily shaped waveguides. To show the accuracy of this method, various examples are solved and the results are compared to those analytical, when there is an analytical solution, and those presented in the literature. The proposed method did not present spurious modes, and in all the examples presented showed excellent results considering a reasonable number of elements in the boundary.     

Kinetic study of gossypol fixation in cottonseed oil

Summary It was shown in experiments carried out under anaerobic conditions that the fixation of gossypol in cottonseed oil is a reaction of the second order with respect to gossypol. In other words, the rate of fixation is proportional to the square of the gossypol concentration in the oil. The rate of fixation is temperature-dependent and increases 22-fold with an increase in temperature from 40° to 80°C. Presented at the 50th Annual Meeting of the American Oil Chemists' Society, New Orleans, La., April 20–22, 1959. One of the laboratories of the Southern Utilization Research and Development Division, Agricultural Research Service, U.S. Department of Agriculture.     

ATMOS技术用于不同等级卫生纸的生产

福伊特造纸的ATMOS技术是卫生纸生产技术的一大突破,它解决了同一台纸机生产不同等级卫生纸的问题,在市场上具有广阔的应用前景。     

Variability of intact rock mechanical properties for some metamorphic rock types and its implications on the number of test specimens

Intact rock strength and stiffness properties are commonly used in rock mass mechanical characterization, and their evaluation is usually based on laboratory tests. Due to the variability that affects strength and stiffness parameters, the determination of the number of laboratory-tested specimens required to obtain a reliable reference value is very useful. However, many studies reported in apposite literature focused only on the variability of strength parameters. This study investigates the variability of some of the most important strength and stiffness properties (unconfined compressive strength, indirect tensile strength, tangent and secant Young’s moduli, Poisson’s ratio) by applying statistical methods (statistical decision theory and statistical inference theory). A data set of 451 laboratory tests was used, performed on three rock types. The statistical analyses were applied with the aim of assessing how closely intact rock laboratory data follow a normal distribution and determining the minimum number of specimens required to obtain a reliable average value of the parameters in relation to a targeted precision index for a confidence level of 95 %. The results indicate that the minimum number of samples needed varies depending on rock and test types. Among the stiffness properties, tangent Young’s modulus has a lower variability than both the secant modulus and the Poisson’s ratio, whereas in terms of strength parameters, unconfined compressive strength is subject to greater variability than indirect tensile strength.     

Hybrid Polyoxometalates: Merging Organic and Inorganic Domains for Enhanced Catalysis and Energy Applications

Organic–inorganic hybrids based on polyoxometalate scaffolds(POMs) are a unique class of molecular metal-oxides featuring a composite surface, whereby the merging of complementary domains stimulates new functions and enhances performances. The interaction between the organic and inorganic components can be designed via covalent and/or non-covalent strategies, yielding novel molecular systems with key applications in catalysis and materials science. Selected examples of such a rewarding approach will be illustrated, including the synthesis of tailored POM-based catalysts, and their application in homogeneous systems and on electrocatalytic surfaces for water splitting and renewable energy production.