Static, Dynamic, and Hybrid Neural Networks in Forecasting Inflation

The back-propagation neural network (BPN) model has been the most popular form of artificial neural network model used for forecasting, particularly in economics and finance. It is a static (feed-forward) model which has a learning process in both hidden and output layers. In this paper we compare the performance of the BPN model with that of two other neural network models, viz., the radial basis function network (RBFN) model and the recurrent neural network (RNN) model, in the context of forecasting inflation. The RBFN model is a hybrid model with a learning process that is much faster than the BPN model and that is able to generate almost the same results as the BPN model. The RNN model is a dynamic model which allows feedback from other layers to the input layer, enabling it to capture the dynamic behavior of the series. The results of the ANN models are also compared with those of the econometric time series models.     

Estimation of Time-Varying Parameters in Statistical Models: An Optimization Approach

We propose a convex optimization approach to solving the nonparametric regression estimation problem when the underlying regression function is Lipschitz continuous. This approach is based on the minimization of the sum of empirical squared errors, subject to the constraints implied by Lipschitz continuity. The resulting optimization problem has a convex objective function and linear constraints, and as a result, is efficiently solvable. The estimated function computed by this technique, is proven to convergeto the underlying regression function uniformly and almost surely, when the sample size grows to infinity, thus providing a very strong form of consistency. Wealso propose a convex optimization approach to the maximum likelihood estimation of unknown parameters in statistical models, where the parameters depend continuously on some observable input variables. For a number of classical distributional forms, the objective function in the underlying optimization problem is convex and the constraints are linear. These problems are, therefore, also efficiently solvable.     

Tertiary-treated municipal wastewater is a significant point source of antibiotic resistance genes into Duluth-Superior Harbor

In this study, the impact of tertiary-treated municipal wastewater on the quantity of several antibiotic resistance determinants in Duluth-Superior Harbor was investigated by collecting surface water and sediment samples from 13 locations in Duluth-Superior Harbor, the St. Louis River, and Lake Superior. Quantitative PCR (qPCR) was used to target three different genes encoding resistance to tetracycline (tet(A), tet(X), and tet(W)), the gene encoding the integrase of class 1 integrons (intI1), and total bacterial abundance (16S rRNA genes) as well as total and human fecal contamination levels (16S rRNA genes specific to the genus Bacteroides ). The quantities of tet(A), tet(X), tet(W), intI1, total Bacteroides , and human-specific Bacteroides were typically 20-fold higher in the tertiary-treated wastewater than in nearby surface water samples. In contrast, the quantities of these genes in the St. Louis River and Lake Superior were typically below detection. Analysis of sequences of tet(W) gene fragments from four different samples collected throughout the study site supported the conclusion that tertiary-treated municipal wastewater is a point source of resistance genes into Duluth-Superior Harbor. This study demonstrates that the discharge of exceptionally treated municipal wastewater can have a statistically significant effect on the quantities of antibiotic resistance genes in otherwise pristine surface waters.     

Accumulation and depuration of polychlorinated biphenyls from field-collected sediment in three freshwater organisms

As laboratory-based bioaccumulation methods are standardized and expanded to include other test species, kinetic studies assessing the major classes of contaminants with these species are needed to adequately select the standard duration for bioaccumulation tests. In the present study we measured the uptake (28-d exposure) of polychlorinated biphenyls (PCBs; total and selected congeners) from field-contaminated sediment in the oligochaete Lumbriculus variegatus, mayfly nymph Hexagenia spp., and fathead minnow Pimephales promelas. Depuration (25 d) of PCBs was measured in organisms that had been transferred to clean sediment after the 28-d exposure. Uptake and elimination of PCBs was rapid in L. variegatus and Hexagenia spp. Tissue residues reached steady-state concentrations within 28 d; elimination rates and biota-sediment accumulation factors (BSAFs) of the PCB congeners were not correlated with K(OW). Uptake and elimination rates of PCBs were slower in P. promelas, and it is not clear whether steady-state was reached in fish tissues. Elimination rates of the PCB congeners significantly decreased with increasing K(OW) in fish. The appropriateness of a 28-d exposure for measuring steady-state concentrations in tissue of the invertebrates was confirmed, but further study is required for fish.     

Metal Dusting of Alumina-Forming Creep-Resistant Austenitic Stainless Steels

Three developmental alumina-forming austenitic stainless steels were exposed to metal dusting conditions at 650?°C in a gas of 50%CO–49%H₂–1%H₂O (a _C: 36.7 and $ p_{{{\text{O}}_{2} }} $ : 2.83?×?10^?26?atm) under thermal cycling conditions. Metal wastage measurement showed initially slow kinetics followed by a fast weight loss. This observation is attributed to the formation of protective chromia/alumina oxide scales in the early stage of the reaction, followed by local oxide failure/spallation during cyclic reaction. Metal dusting initiated from these local defects, and pitting-type attack was observed after 131 cycles of reaction. After 352 cycles, severe dusting had developed, forming heavy and distinctive “tentacles” of superficial coke. This carbon deposit was composed of fine carbon filaments. Examination by TEM of the coke-metal reaction front showed direct surface metal disintegration, indicating that the dusting follows the classical mechanism for austenitic materials. Etching with aqua regia revealed a carburised zone formed in the alloy underneath the coke layer. Analysis by TEM of this zone revealed the formation of ultra-fine, needle-shaped chromium carbide precipitates within a chromium depleted austenite matrix.     

Improving austenitic stainless steel resistance spot weld quality using external magnetic field

A ring-shaped permanent magnet is applied in resistance spot welding to improve the weld quality of austenitic stainless steel. Under the action of an external magnetic field, the profile of the weld nugget became peanut-shaped instead of ellipsoidal. The crystal orientation near the faying surface was less directional, and equiaxed grains were formed in weld nugget centre. Moreover, the shrinkage cavities tendency in traditional resistance spot welds was reduced. The relatively slow cooling speed could cause element segregation and thus change solidification mode and significantly affect weld microstructures. The mechanical performance of the welds was finally improved by the applied external magnetic field in terms of microhardness and lap-shear strength.     

Protein nutrition of dairy cows receiving grass silage diets. Effects on silage intake and milk production of postruminal supplements of casein or soya-protein isolate and the effects of intravenous infusions of a mixture of methionine,phenylalanine and tryptophan

Milk production responses to protein supplementation of grass silage diets were investigated in two experiments with dairy cows. In Experiment 1. four cows received a basal diet of grass silage ad libitum plus a barley/soya supplement at 5 kg day^?1. The cows received the following treatments in a 4 × 4 Latin square design with 10-day periods: (1) basal diet, (2) basal diet plus 230 g casein day^?1 infused into the abomasum; (3) basal diet plus 225 g soya protein isolate (SPI) day^?1 into the abomasum; and (4) basal diet plus a dietary addition of 1.2 kg fishmeal-based product (AP) day^?1. All three supplements increased (P < 0.05) the yield of milk and milk protein relative to the basal diet, the values being 16.9. 535:18.8. 610; 19.5, 625; and 18.5 kg day^?1 and 582 g day^?1 for the basal. AP, casein and SPI treatments respectively. Responses to casein were greater (P < 0.05) than to SPI. Changes in the concentrations of amino acids in blood plasma suggested that methionine, tryptophan and phenylalanine were in shortest supply relative to demand, and their potential role as limiting amino acids was investigated in a second experiment. In Experiment 2, four dairy cows were used in a 4 × 4 Latin square design with 10-day periods. The four treatments were (1) a diet of grass silage ad libitum plus 5 kg day^?1 of a barley/soya supplement containing 176 g CP kg^?1 DM (LP); (2) LP plus an intravenous infusion of 5.0 g methionine, 9.1 g phenylalanine and 2.2 g tryptophan day^?1; (3) grass silage ad libitum plus 5 kg day^?1 of a barley/soya supplement containing 334 g CP kg^?1 DM (HP); and (4) HP plus the intravenous infusion of methionine, phenylalanine and tryptophan. The HP treatment produced a higher yield (P < 0.05) of milk protein than the LP treatment (722 vs. 667 g day^?1) but the amino acid infusion had no effect with either treatment. The elimination of methionine, phenylalanine and tryptophan (and, in previous experiments, methionine and lysine) as candidates for limiting amino acids on diets of grass silage plus soya-based supplements questions whether the differences in responses to casein and SPI can be explained in terms of their amino acid compositions. The role of factors such as peplides, producing during digestion, some of which may be biologically active, deserves investigation.     

Characterization of disinfection byproduct precursors based on hydrophobicity and molecular size

Natural organic matter (NOM) from five water sources was fractionated using XAD resins and ultrafiltration membranes into different groups based on hydrophobicity and molecular weight (MW), respectively. The disinfection byproduct formation from each fraction during chlorination and chloramination was studied. In tests using chlorination, hydrophobic and high MW (e.g., >0.5 kDa) precursors produced more unknown total organic halogen (UTOX) than corresponding hydrophilic and low MW (e.g., <0.5 kDa) precursors. Trihaloacetic acid (THAA) precursors were more hydrophobic than trihalomethane (THM) precursors. The formation of THM and THAA was similar among different fractions for a water with low humic content. Hydrophilic and low MW (<0.5 kDa) NOM fractions gave the highest dihaloacetic acid (DHAA) yields. No significant difference was found for DHAA formation among different NOM fractions during chloramination. Increasing pH from 6 to 9 led to lower TOX formation for hydrophobic and high MW NOM fractions but had little impact on TOX yields from hydrophilic and low MW fractions. Bromine and iodine were more reactive with hydrophilic and low MW precursors as measured by THM or HAA formation than their corresponding hydrophobic and high MW precursors. However, hydrophobic and high MW precursors produced more UTOX when reacting with bromine and iodine.