EnergyPlus is the U.S. Department of Energy's flagship whole-building energy simulation engine and provides extensive simulation capabilities. However, the computational cost of these capabilities has resulted in annual building simulations that typically requires 2-3 min of wall-clock time to complete. While EnergyPlus's overall speed is improving (EnergyPlus 7.0 is 25-40% faster than EnergyPlus 6.0), the overall computational burden still remains and is the top user complaint. In other engineering domains, researchers substitute surrogate or approximate models for the computationally expensive simulations to improve simulation and reduce calibration time. Previous work has successfully demonstrated small-scale EnergyPlus surrogate models that use 10-16 input variables to estimate a single output variable. This work leverages feed forward neural networks and Lasso regression to construct robust large-scale EnergyPlus surrogate models based on 3 benchmark datasets that have 7-156 inputs. These models were able to predict 15-min values for most of the 80-90 simulation outputs deemed most important by domain experts within 5% (whole building energy within 0.07%) and calculate those results within 3 s, greatly reducing the required simulation runtime for relatively close results. The techniques shown here allow any software to be approximated by machine learning in a way that allows one to quantify the trade-off of accuracy for execution time. (C) 2017 Elsevier Ltd. All rights reserved.

Clinical & Experimental AllergyVolume 45, Issue 7 p. 1262-1265 Research Letter Ragweed pollen: is climate change creating a new aeroallergen problem in the UK? C. H. Pashley, Corresponding Author C. H. Pashley Department of Infection, Immunity and Inflammation, Institute for Lung Health, University of Leicester, Leicester, UK Correspondence: Dr Catherine H. Pashley, Aerobiology & Clinical Mycology, Department of Infection, Immunity and Inflammation, University of Leicester, University Road, Leicester LE1 9HN, UK. E-mail: [email protected]Search for more papers by this authorJ. Satchwell, J. Satchwell Department of Infection, Immunity and Inflammation, Institute for Lung Health, University of Leicester, Leicester, UKSearch for more papers by this authorR. E. Edwards, R. E. Edwards Department of Infection, Immunity and Inflammation, Institute for Lung Health, University of Leicester, Leicester, UKSearch for more papers by this author C. H. Pashley, Corresponding Author C. H. Pashley Department of Infection, Immunity and Inflammation, Institute for Lung Health, University of Leicester, Leicester, UK Correspondence: Dr Catherine H. Pashley, Aerobiology & Clinical Mycology, Department of Infection, Immunity and Inflammation, University of Leicester, University Road, Leicester LE1 9HN, UK. E-mail: [email protected]Search for more papers by this authorJ. Satchwell, J. Satchwell Department of Infection, Immunity and Inflammation, Institute for Lung Health, University of Leicester, Leicester, UKSearch for more papers by this authorR. E. Edwards, R. E. Edwards Department of Infection, Immunity and Inflammation, Institute for Lung Health, University of Leicester, Leicester, UKSearch for more papers by this author First published: 22 May 2015 https://doi.org/10.1111/cea.12572Citations: 7Read the full textAboutPDF ToolsRequest permissionExport citationAdd to favoritesTrack citation ShareShare Give accessShare full text accessShare full-text accessPlease review our Terms and Conditions of Use and check box below to share full-text version of article.I have read and accept the Wiley Online Library Terms and Conditions of UseShareable LinkUse the link below to share a full-text version of this article with your friends and colleagues. Learn more.Copy URL Share a linkShare onEmailFacebookTwitterLinkedInRedditWechat References 1Smith M, Cecchi L, Skjoth CA, Karrer G, Sikoparija B. Common ragweed: a threat to environmental health in Europe. Environ Int 2013; 61: 115–26. 2Storkey J, Stratonovitch P, Chapman DS, Vidotto F, Semenov MA. A process-based approach to predicting the effect of climate change on the distribution of an invasive allergenic plant in Europe. PLoS ONE 2014; 9: e88156. 3de Weger LA, Bergmann KC, Rantio-Lehtimäki A et al. Impact of pollen. In: M Sofiev, K-C Bergmann, eds. Allergenic Pollen: a review of the production, release, distribution and health impacts. Dordrecht: Springer, 2013: 161–215. 4Burbach GJ, Heinzerling LM, Rohnelt C, Bergmann KC, Behrendt H, Zuberbier T, study G. Ragweed sensitization in Europe – GA(2)LEN study suggests increasing prevalence. Allergy 2009; 64: 664–5. 5Pashley CH, Fairs A, Edwards RE, Bailey JP, Corden JM, Wardlaw AJ. Reproducibility between counts of airborne allergenic pollen from two cities in the East Midlands, UK. Aerobiologia 2009; 25: 249–63. 6D'Amato G, Cecchi L, Bonini S et al. Allergenic pollen and pollen allergy in Europe. Allergy 2007; 62: 976–90. 7Asero R. Ragweed allergy in northern Italy: are patterns of sensitization changing? Eur Ann Allergy Clin Immunol 2012; 44: 157–9. 8Asero R, Wopfner N, Gruber P, Gadermaier G, Ferreira F. Artemisia and Ambrosia hypersensitivity: co-sensitization or co-recognition? Clini Exp Allergy 2006; 36: 658–65. 9Smith M, Jaeger S, Berger U et al. Geographic and temporal variations in pollen exposure across Europe. Allergy 2014; 69: 913–23. Citing Literature Volume45, Issue7July 2015Pages 1262-1265 ReferencesRelatedInformation

Building energy modeling BEM is an approach to model the energy usage in buildings for design and retrofit purposes. EnergyPlus is the flagship Department of Energy software that performs BEM for different types of buildings. The input to EnergyPlus can often extend in the order of a few thousand parameters that have to be calibrated manually by an expert for realistic energy modeling. This makes it challenging and expensive thereby making BEM unfeasible for smaller projects. In this paper, we describe the 'Autotune' research that employs machine learning algorithms to generate agents for the different kinds of standard reference buildings in the US building stock. The parametric space and the variety of building locations and types make this a challenging computational problem necessitating the use of supercomputers. Millions of EnergyPlus simulations are run on supercomputers that are subsequently used to train machine learning algorithms to generate agents. These agents, once created, can then run in a fraction of the time thereby allowing cost-effective calibration of building models. Published 2014. This article is a US Government work and is in the public domain in the USA.

Many key building design policies are made using sophisticated computer simulations such as EnergyPlus (E+), the DOE flagship whole-building energy simulation engine. E+ and other sophisticated computer simulations have several major problems. The two main issues are 1) gaps between the simulation model and the actual structure, and 2) limitations of the modeling engine's capabilities. Currently, these problems are addressed by having an engineer manually calibrate simulation parameters to real world data or using algorithmic optimization methods to adjust the building parameters. However, some simulations engines, like E+, are computationally expensive, which makes repeatedly evaluating the simulation engine costly. This work explores addressing this issue by automatically discovering the simulation's internal input and output dependencies from ~20 Gigabytes of E+ simulation data, future extensions will use ~200 Terabytes of E+ simulation data. The model is validated by inferring building parameters for E+ simulations with ground truth building parameters. Our results indicate that the model accurately represents parameter means with some deviation from the means, but does not support inferring parameter values that exist on the distribution's tail.

Kernel methods have difficulties scaling to large modern data sets. The scalability issues are based on computational and memory requirements for working with a large matrix. These requirements have been addressed over the years by using low-rank kernel approximations or by improving the solvers' scalability. However, Least Squares Support Vector Machines (LS-SVM), a popular SVM variant, and Kernel Ridge Regression still have several scalability issues. In particular, the O(n^3) computational complexity for solving a single model, and the overall computational complexity associated with tuning hyper parameters are still major problems. We address these problems by introducing an O(nlog n) approximate l-fold cross-validation method that uses a multi-level circulant matrix to approximate the kernel. In addition, we prove our algorithm's computational complexity and present empirical runtimes on data sets with approximately one million data points. We also validate our approximate method's effectiveness at selecting hyper parameters on real world and standard benchmark data sets. Lastly, we provide experimental results on using a multi level circulant kernel approximation to solve LS-SVM problems with hyper parameters selected using our method.

Building Energy Modeling (BEM) is an approach to model the energy usage in buildings for design and retrofit purposes. EnergyPlus is the flagship Department of Energy software that performs BEM for different types of buildings. The input to EnergyPlus can often extend in the order of a few thousand parameters which have to be calibrated manually by an expert for realistic energy modeling. This makes it challenging and expensive thereby making building energy modeling unfeasible for smaller projects. In this paper, we describe the "Autotune" research which employs machine learning algorithms to generate agents for the different kinds of standard reference buildings in the U.S. building stock. The parametric space and the variety of building locations and types make this a challenging computational problem necessitating the use of supercomputers. Millions of EnergyPlus simulations are run on supercomputers which are subsequently used to train machine learning algorithms to generate agents. These agents, once created, can then run in a fraction of the time thereby allowing cost-effective calibration of building models.

Traditional whole building energy modeling suffers from several factors, including the large number of inputs required for building characterization, simplifying assumptions, and the gap between the as-designed and as-built building. Prior work has attempted to mitigate these problems by using sensor-based machine learning approaches to statistically model energy consumption, applying the techniques primarily to commercial building data, which makes use of hourly consumption data. It is unclear, however, whether these techniques can translate to residential buildings, since the energy usage patterns may vary significantly. Until now, most residential modeling research only had access to monthly electrical consumption data. In this article, we report on the evaluation of seven different machine learning algorithms applied to a new residential data set that contains sensor measurements collected every 15 min, with the objective of determining which techniques are most successful for predicting next hour residential building consumption. We first validate each learner's correctness on the ASHRAE Great Energy Prediction Shootout, confirming existing conclusions that Neural Network-based methods perform best on commercial buildings. However, our additional results show that these methods perform poorly on residential data, and that Least Squares Support Vector Machines perform best - a technique not previously applied to this domain. (C) 2012 Elsevier B.V. All rights reserved.

Probabilistic latent semantic analysis (PLSA) has been widely used in the machine learning community. However, the original PLSAs are not capable of modeling real-valued observations and usually have severe problems with over fitting. To address both issues, we propose a novel, regularized Gaussian PLSA (RG-PLSA) model that combines Gaussian PLSAs and hierarchical Gaussian mixture models (HGMM). We evaluate our model on supervised human action recognition tasks, using two publicly available datasets. Average classification accuracies of 97.69% and 93.72% are achieved on the Weizmann and KTH Action Datasets, respectively, which demonstrate that the RG-PLSA model outperforms Gaussian PLSAs and HGMMs, and is comparable to the state of the art.

Metabolic profiling of biofluids, based on the quantitative analysis of the concentration profile of their free low molecular mass metabolites, has been playing increasing role employed as a means to gain understanding of the progression of metabolic disorders, including obesity. Chromatographic methods coupled with mass spectrometry have been established as a strategy for metabolic profiling. Among these, GC–MS, targeting mainly the primary metabolism intermediates, offers high sensitivity, good peak resolution and extensive databases. However, the derivatization step required for many involatile metabolites necessitates specific data validation, normalization and analysis protocols to ensure accurate and reproducible performance. In this study, the GC–MS metabolic profiles of plasma samples from mice maintained on 12- or 15-month long low (10kcal%) or high (60kcal%) fat diets were obtained. The profiles of the trimethylsilyl(TMS)-methoxime(MeOx) derivatives of the free polar metabolites were acquired through GC–(ion trap)MS, using [U–13C]-glucose as the internal standard. After the application of a recently developed data correction and normalization/filtering protocol for GC–MS metabolomic datasets, the profiles of 48 out of the 77 detected metabolites were used in multivariate statistical analysis. Data mining suggested a decrease in the activity of the energy metabolism with age. In addition, the metabolic profiles indicated the presence of subpopulations with different physiology within the high- and low-fat diet mice, which correlated well with the difference in body weight among the animals and current knowledge about hyperglycemic conditions.

In this paper we present Myro-C++, developed at the University of Tennessee. Myro-C++ is a C++ port of the Python Myro library that was written by the Institute for Personal Robots in Education (IPRE) at Georgia Tech and Bryn Mawr College. Myro-C++ is publicly available, open source software, released under the GPLv3 open source license. At the time of writing, the library has been used six semesters for the CS1 course at the University of Tennessee, Knoxville. The library contains functions for control of the robot and access to sensor information, and provides the ability to display the live camera image from the robot into a video window. This library is used as a teaching tool in our CS1 course where students learn basic programming fundamentals using multiple artificial intelligence based labs. In addition to the software, the IPRE book, Learning Computing with Robots, has been edited to use C++ examples and explanations, and is freely available. We also present example programs that we use as laboratory assignments in our Introduction to Computer Science course, which are also freely available.

There is a growing interest in building Internet-scale sensor networks that integrate sensors from around the world into a single unified system. In contrast, robotics application development has primarily focused on building specialized systems. These specialized systems take scalability and reliability into consideration, but generally neglect exploring the key components required to build a large scale system. Integrating robotic applications with Internet-scale sensor networks will unify specialized robotics applications and provide answers to large scale implementation concerns. We focus on utilizing Internet-scale sensor network technology to construct a framework for unifying robotic systems. Our framework web-enables a surveillance robot's sensor observations and provides a web-interface to the robot's actuators. This lets robots seamlessly integrate into web applications. In addition, the framework eliminates most prerequisite robotics knowledge, allowing for the creation of general web-based robotics applications. The framework also provides mechanisms to create applications that can interface with any robot. Frameworks such as this one are key to solving large scale mobile robotics implementation problems. We provide an overview of previous Internet-scale sensor networks, Sensorpedia (an ad-hoc Internet-scale sensor network), our framework for integrating robots with Sensorpedia, two applications which illustrate our frameworks ability to support general web-based robotic control, and offer experimental results that illustrate our framework's scalability, feasibility, and resource requirements.

Historically in the East Midlands, UK, airborne pollen has been monitored in two cities, Derby and Leicester, situated 41 km (25 miles) apart. The aim of the present study was to compare aerobiological data from both sites to determine if a forecast based on data from one site would be sufficient for both, and to address the wider issue of reproducibility between geographically separated sites. Pollen types recorded could be split into two groups according to annual abundance, maximum daily concentration and the number of high count days. Six taxa made up the abundant group; ash, birch, grass, oak, nettle-type and yew-type, representing 90 and 88% of the total air spora for Derby and Leicester, respectively. Three consecutive years of grass and nettle pollen data are presented, supported by one year of abundant tree pollen data. There were highly significant positive correlations between the counts obtained. Line charts showing the average number of pollen grains m −3 air day −1 show similar trends, and Bland–Altman plots show little discrepancy between the amounts of pollen counted on any given day. Each day was classified according to the UK accepted threshold levels for grass. Weighted kappa statistics showed substantial or almost perfect agreement between the forecast classifications. With the caveat that this would not apply in a region with restrictions to air flow such as a mountain range or with extreme fluctuations such as a coastline site, this study suggests that data from a single site is suitable for forecasting a distance of up to 41 km.

BACKGROUND. The TRansgenic Adenocarcinoma of the Mouse Prostate (TRAMP) mouse model has frequently been used in preclinical studies with chemotherapeutic/chemopreventive rationales. Here the hypothesis was tested using H-1-NMR-based metabolic profiling that the TRAMP tumor metabolic phenotype resembles that reported for human prostate cancer.METHODS. Aqueous extracts or intact tissues of normal prostate from 8- ("young") or 28-("old") week-old C57BL/6J wild-type mice or of prostate tumor from age-matched TRAMP mice were analyzed by H-1-NMR. Results were compared with immunohistochemical findings. Expression of choline kinase was studied at the protein and mRNA levels.RESULTS. In young TRAMP mice presenting with zonal hyperplasia, the ratio of glycerophosphocholine (GPC) to phosphocholine (PC) was 22% below that in wild-type mice (P < 0.05). In old TRAMP mice with well-defined malignancy, reduced tumor levels of citrate (49%), choline (33%), PC (57%), GPC (66%), and glycerophosphoinositol (61%) were observed relative to normal prostate (P < 0.05). Hierarchical cluster analysis of metabolite levels distinguished between normal and malignant tissue in old but not young mice. While the reduction in tissue citrate resembles human prostate cancer, low levels of choline species in TRAMP tumors suggest atypical phospholipid metabolism as compared to human prostate cancer. TRAMP tumor and normal prostate tissues did not differ in expression of choline kinase, which is overexpressed in human prostate cancer.CONCLUSION. Although prostate cancer in TRAMP mice shares some metabolic features with that in humans, it differs with respect to choline phospholipid metabolism, which could impact upon the interpretation of results from biomarker or chemotherapy/chemoprevention studies.

The dysfunction of hepatic heme synthesis by 2,3,7,8-tetrachlordibenzo- p-dioxin (TCDD) in mice, enhanced by iron, leads to accumulation of uroporphyrins I and III (uroporphyria) and resembles the human disorder porphyria cutanea tarda (PCT) precipitated by alcohol and estrogenic drugs. Although consequences of TCDD are considered entirely dependent on the aryl hydrocarbon receptor (AHR), this is not proven for uroporphyria. Administration of TCDD (75 microg/kg) caused uroporphyria in susceptible C57BL/6J mice with high-affinity AHR after 5 weeks (>600-fold increase in hepatic uroporphyrins). Transcriptomics showed significant modified gene expressions for intermediary, heme, and iron metabolism as well as for oxidative stress and cell injury. Resistant low-affinity AHR DBA/2 mice (no increase in porphyrins) showed far fewer changes. At this dose of TCDD, persistent up-regulation of some traditional AH battery genes occurred in both strains. Essentiality of AHR was demonstrated with C57BL/6 Ahr knockout mice. Elevation of hepatic uroporphyrins was 964-fold in Ahr (+/+) mice, lower in Ahr (+/-) (60-fold), but undetectable with Ahr (-/-) . Consistent with an oxidative mechanism, iron overload enhanced porphyria as well as general liver injury in Ahr (+/+) and Ahr (+/-) mice but had no interactive effect in Ahr (-/-) . In contrast, when iron-treated mice received, instead of TCDD, the heme precursor 5-aminolevulinic acid (ALA), causing uroporphyia in Ahr (+/+) mice (242-fold rise in uroporphyrins), elevation of uroporphyrins I and III (42-fold) also occurred in Ahr (-/-) mice and was seemingly associated with AHR-independent expression of Cyp1a2. The findings prove that AHR is a key factor in porphyria induced in mice by TCDD. However, in other models of human PCT, participation of AHR may not be an essential requirement.

Tea consumption is associated with a reduced risk of mammary cancer as reflected by epidemiological studies and experiments in carcinogen-induced rodent models of mammary carcinogenesis. We tested the hypothesis that green tea catechins (GTC) or theaflavins from black tea (BTT) interfere with mammary carcinogenesis in C3(1) SV40 T,t antigen transgenic multiple mammary adenocarcinoma (TAg) mice and that GTC/BTT affect tumor survival or oxidation status. TAg mice received GTC/BTT (0.05%) in drinking water for their lifetime. As compared to control mice, they survived longer and had smaller tumors. On microscopic inspection, the size of the largest tumor per mouse was decreased by 40-42% (p < 0.01). GTC (0.01%) and BTT (0.05%) increased levels of cleaved caspase 3 in tumor tissue by 67 and 38%, respectively (p < 0.05), intimating increased apoptosis. Tumor levels of the malondialdehyde-DNA adduct M(1)dG in mice receiving GTC or BTT (0.05%) were reduced by 78 (p < 0.001) or 63% (p < 0.05), respectively, as compared to controls. The results render the exploration of the breast cancer chemopreventive properties of tea preparations in humans worthwhile.

The effects of subcutaneous dosing of neonatal CD-1 mice with tamoxifen on days 1-5 after birth at doses of 0, 5, 10, 25 or 50 microg/pup or with 4-hydroxyoestradiol at 2 microg/pup have been investigated. Animals were culled at 1.5, 3, 6, 12 and 18 months after dosing and changes in uterine and ovarian pathology examined. Results showed both compounds to result in uterine hypoplasia relative to controls. At 18 months after dosing in the uterus, there was a fairly marked atrophy of the muscle layer, mild to moderate glandular hyperplasia of the endometrium even though these irregularly shaped glands did not penetrate through the myometrium and no adenocarcinomas were detected. At 18 months after dosing, oviducts showed mild focal adenomatous changes characterized by penetration epithelial hyperplasia, changes similar to those previously reported as 'diverticulosis and salpingitis isthmica nodosa' following diethylstilbestrol treatment of mice. At this time, both tamoxifen and 4-hydroxyoestradiol also affected the ovaries which showed a paucity of follicles and no corpora lutea, suggesting that there had been disruption to the oestrus cycle, particularly with tamoxifen at the highest dose where the ovaries of mice contained no developing follicles. At 18 months, control mice were cycling normally. Results failed to substantiate that tamoxifen and 4-hydroxyoestradiol are uterine carcinogens in this neonatal mouse model.

Polymorphisms of genes linked to iron metabolism may account for individual variability in hemochromatosis and iron status connected with liver and cardiovascular diseases, cancers, toxicity, and infection. Mouse strains exhibit marked differences in levels of non‐heme iron, with C57BL/6J and SWR showing low and high levels, respectively. The genetic basis for this variability was examined using quantitative trait loci (QTL) analysis together with expression profiling and chromosomal positions of known iron‐related genes. Non‐heme iron levels in liver and spleen of C57BL/6J × SWR F2 mice were poorly correlated, indicating independent regulation. Highly significant (P < .01) polymorphic loci were found on chromosomes 2 and 16 for liver and on chromosomes 8 and 9 for spleen. With sex as a covariate, additional significant or suggestive (P < 0.1) QTL were detected on chromosomes 7, 8, 11, and 19 for liver and on chromosome 2 for spleen. A gene array showed no clear association between most loci and differential iron‐related gene expression. The gene for transferrin and a transferrin‐like gene map close to the QTL on chromosome 9. Transferrin saturation was significantly lower in C57BL/6J mice than in SWR mice, but there was no significant difference in the serum level of transferrin, hepatic expression, or functional change in cDNA sequence. β2‐Microglobulin, which, unlike other loci, was associated with C57BL/6J alleles, is a candidate for the chromosome 2 QTL for higher iron. In conclusion, the findings show the location of polymorphic genes that determine basal iron status in wild‐type mice. Human equivalents may be pertinent in predisposition to hepatic and other disorders. (HEPATOLOGY 2006;44:174–185.)