Information Theory and Neural Networks for Managing Uncertainty in Flood Routing

This paper presents an approach for handling uncertainties arising mainly from ignored or misrepresented processes in physically based models. The approach is based on the application of a parallel artificial neural network (ANN) model that uses state variables, input and output data, and previous model errors at specific time steps to predict the errors of a physically based model. Concepts from information theory are used to discover the relationships between the variables and the model errors, which also serves as a mechanism to detect the predictability of the errors. The resulting information is used to select the best related input data for the error prediction model. The error prediction model is then trained and applied to improve the forecasts made by the physically based model. This approach was applied to a routing model of a 70 km reach of the River Wye, United Kingdom. The results demonstrate that errors from the physically based model show a consistent trend governed by some dynamics of their own, which can be modeled with learning algorithms. Errors were forecasted at different lead times. In all cases the forecasts made by the combined application of both models were more accurate than those made by the physically based model alone. From this it was concluded that, along with proper information analysis techniques, the use of ANN models to predict the forecast errors of physically based models can help to improve significantly the prediction and therefore to reduce the associated uncertainty.     

An empirical study of software release notes

Release notes are an important source of information about a new software release. Such notes contain information regarding what is new, changed, and/or got fixed in a release. Despite the importance of release notes, they are rarely explored in the research literature. Little is known about the contained information, e.g., contents and structure, in release notes. To better understand the types of contained information in release notes, we manually analyzed 85 release notes across 15 different software systems. In our manual analysis, we identify six different types of information (e.g., caveats and addressed issues) that are contained in release notes. Addressed issues refer to new features, bugs, and improvements that were integrated in that particular release. We observe that most release notes list only a selected number of addressed issues (i.e., 6-26 % of all addressed issues in a release). We investigated nine different factors (e.g., issue priority and type) to better understand the likelihood of an issue being listed in release notes. The investigation is conducted on eight release notes of three software systems using four machine learning techniques. Results show that certain factors, e.g., issue type, have higher influence on the likelihood of an issue to be listed in release notes. We use machine learning techniques to automatically suggest the issues to be listed in release notes. Our results show that issues listed in all release notes can be automatically determined with an average precision of 84 % and an average recall of 90 %. To train and build the classification models, we also explored three scenarios: (a) having the user label some issues for a release and automatically suggest the remaining issues for that particular release, (b) using the previous release notes for the same software system, and (c) using prior releases for the current software system and the rest of the studied software systems. Our results show that the content of release notes vary between software systems and across the versions of the same software system. Nevertheless, automated techniques can provide reasonable support to the writers of such notes with little training data. Our study provides developers with empirically-supported advice about release notes instead of simply relying on adhoc advice from on-line inquiries.     

Effects of a titanate coupling agent on the mechanical and thermo‐physical properties of talc‐reinforced polyethylene compounds

An experimental study was carried out to investigate the effects of a titanate coupling agent on the mechanical properties, moisture absorption, and thermal conductivity of talc‐filled high‐density polyethylene (HDPE). Talc (0–35 wt %) was used as reinforcement particulate filler in an HDPE matrix and samples were prepared in a micro‐compounder and an injection molding machine. Isopropyl tri(dioctyl)phosphate titanate (0.5 wt %) was used as coupling agent. Composites with and without coupling agent were evaluated for changes in mechanical and thermo‐physical properties, morphology, and void content. Addition of the titanate coupling agent most often resulted in an increase in stiffness and tensile strength. Furthermore, both the void content and the elongation at break of composites were reduced. Results also showed that the coupling agent had no effects on the thermal conductivity, thermal diffusivity, and specific heat capacity of the composites. In addition, it was observed that the coupling agent was more effective at low concentrations of filler. © 2014 Wiley Periodicals, Inc. J. Appl. Polym. Sci. 2014 , 131, 40449.     

Loss mechanisms in single-mode fiber tapers

The onset of transmission loss in four single-mode tapers is shown to coincide with the taper slope exceeding the critical slope angle at which fundamental mode cutoff occurs. This implies that the primary loss mechanism in single-mode fiber tapers is cutoff of the fundamental mode from the core. A secondary loss mechanism of coupling to the modes of a coaxial outer waveguide is also observed and modeled. Implications for fused fiber couplers are discussed.     

Current-voltage characteristics from an asymptotic analysis of the MOSFET equations

Phenomenological formulae for the current-voltage characteristics of a MOSFET are derived by extending the asymptotic method of Ward. Practical methods for combining the perturbation approximations and numerically implementing the Ward equations are developed. A detailed comparison with real MOSFET data is presented and the model is shown to be effective over a range of device geometries.     

Assessment of iron bioavailability from twenty elite late‐maturing tropical maize varieties using an in vitro digestion/Caco‐2 cell model

An in vitro digestion/Caco‐2 cell model was used to assess iron bioavailability of twenty elite late‐maturing tropical maize varieties grown in three diverse agroecologies in West and Central Africa (WCA). Kernel‐iron concentration of the varieties, averaged across locations, varied from 19.2 to 24.4 mg kg^?1, while mean kernel‐zinc concentration ranged between 19.4 and 24.6 mg kg^?1. Significant differences in iron bioavailability were observed among varieties, but the environment had no significant effect. Mean bioavailable iron ranged between 14% below and 43% above the reference control variety, TZB‐SR. Variety DMR‐LSR‐Y with the highest concentrations of kernel‐iron and ‐zinc of 24–25 mg kg^?1 across the three locations had a similar quantity of bioavailable iron as the reference control, TZB‐SR. In the long run this variety could be potentially effective in reducing iron deficiency because of its high kernel‐iron. The most promising varieties were Mid‐altitude STR synthetic and ACR91SUWAN‐1‐SRC1. They had kernel‐iron and ‐zinc levels of 22–24 mg kg^?1 and bioavailable iron 24–36% higher than the reference control, TZB‐SR. Additional research is necessary to determine if the increases in kernel‐iron concentration and bioavailable iron observed in this study can significantly improve the iron status of individuals in WCA at risk for iron deficiency. Copyright © 2004 Society of Chemical Industry     

Maternal regulation of embryonic growth: the role of vasoactive intestinal peptide

Vasoactive intestinal peptide (VIP) is an important growth regulator of the embryonic day (E)9-E11 mouse. In comparably aged rat embryos, VIP messenger RNA (mRNA) is not detectable; however, peak concentrations of VIP in maternal rat serum indicate a nonembryonic source. In the current study, mouse maternal and embryonic tissues were examined from E6-E12. Although RT-PCR revealed VIP mRNA in E6-E7 conceptuses, by E8 (when extraembryonic tissues could be separated from the embryo), VIP mRNA was detected only in the decidua/trophoblast. Decidual/trophoblastic VIP mRNA decreased until E10, after which it was not detectable. VIP mRNA was not apparent in the embryo until E11-E12. At E9, VIP immunoreactivity was localized to abundant, diffuse cells in the decidua basalis, which were also immunoreactive for T cell markers. VIP binding sites were dense in the decidua/trophoblast at E6, but gradually decreased until E10, after which they were not apparent. VIP binding sites were detected in embryonic neuroepithelium by E9. The transient presence of VIP binding sites and mRNA in the decidua/trophoblast correlate with the critical period of VIP growth regulation, when VIP mRNA is absent in the embryo. These findings suggest that maternal lymphocytes are the source of VIP's regulating early postimplantation embryonic growth.