Strategies for Energy-Efficient Resource Management of Hybrid Programming Models

Many scientific applications are programmed using hybrid programming models that use both message passing and shared memory, due to the increasing prevalence of large-scale systems with multicore, multisocket nodes. Previous work has shown that energy efficiency can be improved using software-controlled execution schemes that consider both the programming model and the power-aware execution capabilities of the system. However, such approaches have focused on identifying optimal resource utilization for one programming model, either shared memory or message passing, in isolation. The potential solution space, thus the challenge, increases substantially when optimizing hybrid models since the possible resource configurations increase exponentially. Nonetheless, with the accelerating adoption of hybrid programming models, we increasingly need improved energy efficiency in hybrid parallel applications on large-scale systems. In this work, we present new software-controlled execution schemes that consider the effects of dynamic concurrency throttling (DCT) and dynamic voltage and frequency scaling (DVFS) in the context of hybrid programming models. Specifically, we present predictive models and novel algorithms based on statistical analysis that anticipate application power and time requirements under different concurrency and frequency configurations. We apply our models and methods to the NPB MZ benchmarks and selected applications from the ASC Sequoia codes. Overall, we achieve substantial energy savings (8.74 percent on average and up to 13.8 percent) with some performance gain (up to 7.5 percent) or negligible performance loss.

Efficient Architectural Design Space Exploration via Predictive Modeling

Efficiently exploring exponential-size architectural design spaces with many interacting parameters remains an open problem: the sheer number of experiments required renders detailed simulation intractable.We attack this via an automated approach that builds accurate predictive models. We simulate sampled points, using results to teach our models the function describing relationships among design parameters. The models can be queried and are very fast, enabling efficient design tradeoff discovery. We validate our approach via two uniprocessor sensitivity studies, predicting IPC with only 1–2% error. In an experimental study using the approach, training on 1% of a 250-K-point CMP design space allows our models to predict performance with only 4–5% error. Our predictive modeling combines well with techniques that reduce the time taken by each simulation experiment, achieving net time savings of three-four orders of magnitude.

Intelligent predictive control for thermal comfort and energy savings in public buildings

This talk addresses the problem of controlling a heating ventilating and air conditioning system with the purpose of achieving a desired thermal comfort level and energy savings. The formulation uses the thermal comfort, assessed using the predicted mean vote (PMV) index, as a restriction and minimises the energy spent to comply with it. This results in the maintenance of thermal comfort and on the minimisation of energy, which in most operating conditions are conflicting goals requiring some sort of optimisation method to find appropriate solutions over time. In this work a discrete model based predictive control methodology is applied to the problem. It consists of three major components: the predictive models, implemented by radial basis function neural networks identifed by means of a multi-objective genetic algorithm [1]; the cost function that will be optimised to minimise energy consumption and provide adequate thermal comfort; and finally the optimisation method, in this case a discrete branch and bound approach. Each component will be described, with a special emphasis on a fast and accurate computation of the PMV indices [2]. Experimental results obtained within different rooms in a building of the University of Algarve will be presented, both in summer [3] and winter [4] conditions, demonstrating the feasibility and performance of the approach. Energy savings resulting from the application of the method are estimated to be greater than 50%.

Use and abuse of mathematical models: an illustration from the 2001 foot and mouth disease epidemic in the United Kingdom

Foot and mouth disease (FMD) is a major threat, not only to countries whose economies rely on agricultural exports, but also to industrialised countries that maintain a healthy domestic livestock industry by eliminating major infectious diseases from their livestock populations. Traditional methods of controlling diseases such as FMD require the rapid detection and slaughter of infected animals, and any susceptible animals with which they may have been in contact, either directly or indirectly. During the 2001 epidemic of FMD in the United Kingdom (UK), this approach was supplemented by a culling policy driven by unvalidated predictive models. The epidemic and its control resulted in the death of approximately ten million animals, public disgust with the magnitude of the slaughter, and political resolve to adopt alternative options, notably including vaccination, to control any future epidemics. The UK experience provides a salutary warning of how models can be abused in the interests of scientific opportunism.

Evaluation of a predictive model for Mycosphaerella graminicola for economic and environmental benefits

A method was developed to evaluate crop disease predictive models for their economic and environmental benefits. Benefits were quantified as the value of a prediction measured by costs saved and fungicide dose saved. The value of prediction was defined as the net gain made by using predictions, measured as the difference between a scenario where predictions are available and used and a scenario without prediction. Comparable 'with' and 'without' scenarios were created with the use of risk levels. These risk levels were derived from a probability distribution fitted to observed disease severities. These distributions were used to calculate the probability that a certain disease induced economic loss was incurred. The method was exemplified by using it to evaluate a model developed for Mycosphaerella graminicola risk prediction. Based on the value of prediction, the tested model may have economic and environmental benefits to growers if used to guide treatment decisions on resistant cultivars. It is shown that the value of prediction measured by fungicide dose saved and costs saved is constant with the risk level. The model could also be used to evaluate similar crop disease predictive models.

Use of a predictive model for food insecurity estimates in Brazil

In 2004 the National Household Survey (Pesquisa Nacional par Amostras de Domicilios - PNAD) estimated the prevalence of food and nutrition insecurity in Brazil. However, PNAD data cannot be disaggregated at the municipal level. The objective of this study was to build a statistical model to predict severe food insecurity for Brazilian municipalities based on the PNAD dataset. Exclusion criteria were: incomplete food security data (19.30%); informants younger than 18 years old (0.07%); collective households (0.05%); households headed by indigenous persons (0.19%). The modeling was carried out in three stages, beginning with the selection of variables related to food insecurity using univariate logistic regression. The variables chosen to construct the municipal estimates were selected from those included in PNAD as well as the 2000 Census. Multivariate logistic regression was then initiated, removing the non-significant variables with odds ratios adjusted by multiple logistic regression. The Wald Test was applied to check the significance of the coefficients in the logistic equation. The final model included the variables: per capita income; years of schooling; race and gender of the household head; urban or rural residence; access to public water supply; presence of children; total number of household inhabitants and state of residence. The adequacy of the model was tested using the Hosmer-Lemeshow test (p=0.561) and ROC curve (area=0.823). Tests indicated that the model has strong predictive power and can be used to determine household food insecurity in Brazilian municipalities, suggesting that similar predictive models may be useful tools in other Latin American countries.

Fuzzy rule base assessment models : theoretical analyses and a case study

An assessment model is usually a mathematical model that produces a measuring index, in the form of a numerical score to a situation/object, with respect to the subject of measure. To allow a valid and useful comparison among various situations/objects according to their associated numerical scores to be made, two important properties, i.e., the monotone output property and output resolution properties, are essential in fuzzy inference-based assessment problems. In this paper, the conditions for a fuzzy assessment model to fulfill the monotone output property is investigated using a derivative approach. A guideline on how the input membership functions should be tuned is also provided. Besides, the output resolution property is defined as the derivative of the output of the assessment model with respect to the input, whereby the derivative should be greater than a minimum resolution. Based on the derivative, improvements to the output resolution property by refining the fuzzy production rules are suggested. A case study on the Bowles fuzzy RPN model to demonstrate the effectiveness of the properties is also included.

Analysis of ensemble models in the medium term hydropower scheduling

The medium term hydropower scheduling (MTHS) problem involves an attempt to determine, for each time stage of the planning period, the amount of generation at each hydro plant which will maximize the expected future benefits throughout the planning period, while respecting plant operational constraints. Besides, it is important to emphasize that this decision-making has been done based mainly on inflow earliness knowledge. To perform the forecast of a determinate basin, it is possible to use some intelligent computational approaches. In this paper one considers the Dynamic Programming (DP) with the inflows given by their average values, thus turning the problem into a deterministic one which the solution can be obtained by deterministic DP (DDP). The performance of the DDP technique in the MTHS problem was assessed by simulation using the ensemble prediction models. Features and sensitivities of these models are discussed. © 2012 IEEE.

Predicting biological parameters of estuarine benthic communities using models based on environmental data

Este trabalho objetivou predizer parâmetros da estrutura de associações macrobentônicas (composição específica, abundância, riqueza, diversidade e equitatividade) em estuários do Sul do Brasil, utilizando modelos baseados em dados ambientais (características dos sedimentos, salinidade, temperaturas do ar e da água, e profundidade). As amostragens foram realizadas sazonalmente em cinco estuários entre o inverno de 1996 e o verão de 1998. Em cada estuário as amostras foram coletadas em áreas não poluídas, com características semelhantes quanto a presença ou ausência de vegetação, profundidade e distância da desenbocadura. Para a obtenção dos modelos de predição, foram utilizados dois métodos: o primeiro baseado em Análise Discriminante Múltipla (ADM) e o segundo em Regressão Linear Múltipla (RLM). Os modelos baseados em ADM apresentaram resultados melhores do que os baseados em regressão linear. Os melhores resultados usando RLM foram obtidos para diversidade e riqueza. É possível então, concluir que modelos como aqui derivados podem representar ferramentas muito úteis em estudos de monitoramento ambiental em estuários.

Risk assessment for invasive exotic plants using multi-criteria risk models

Invasive exotic plants have altered natural ecosystems across much of North America. In the Midwest, the presence of invasive plants is increasing rapidly, causing changes in ecosystem patterns and processes. Early detection has become a key component in invasive plant management and in the detection of ecosystem change. Risk assessment through predictive modeling has been a useful resource for monitoring and assisting with treatment decisions for invasive plants. Predictive models were developed to assist with early detection of ten target invasive plants in the Great Lakes Network of the National Park Service and for garlic mustard throughout the Upper Peninsula of Michigan. These multi-criteria risk models utilize geographic information system (GIS) data to predict the areas at highest risk for three phases of invasion: introduction, establishment, and spread. An accuracy assessment of the models for the ten target plants in the Great Lakes Network showed an average overall accuracy of 86.3%. The model developed for garlic mustard in the Upper Peninsula resulted in an accuracy of 99.0%. Used as one of many resources, the risk maps created from the model outputs will assist with the detection of ecosystem change, the monitoring of plant invasions, and the management of invasive plants through prioritized control efforts.

Sediment organic matter, grain size, and results of prediction models from northern North Atlantic and Arctic Seas

Sediment samples and hydrographic conditions were studied at 28 stations around Iceland. At these sites, Conductivity-Temperature-Depth (CTD) casts were conducted to collect hydrographic data and multicorer casts were conductd to collect data on sediment characteristics including grain size distribution, carbon and nitrogen concentration, and chloroplastic pigment concentration. A total of 14 environmental predictors were used to model sediment characteristics around Iceland on regional geographic space. For these, two approaches were used: Multivariate Adaptation Regression Splines (MARS) and randomForest regression models. RandomForest outperformed MARS in predicting grain size distribution. MARS models had a greater tendency to over- and underpredict sediment values in areas outside the environmental envelope defined by the training dataset. We provide first GIS layers on sediment characteristics around Iceland, that can be used as predictors in future models. Although models performed well, more samples, especially from the shelf areas, will be needed to improve the models in future.

Autonomous classification models in ubiquitous environments

Stream-mining approach is defined as a set of cutting-edge techniques designed to process streams of data in real time, in order to extract knowledge. In the particular case of classification, stream-mining has to adapt its behaviour to the volatile underlying data distributions, what has been called concept drift. Moreover, it is important to note that concept drift may lead to situations where predictive models become invalid and have therefore to be updated to represent the actual concepts that data poses. In this context, there is a specific type of concept drift, known as recurrent concept drift, where the concepts represented by data have already appeared in the past. In those cases the learning process could be saved or at least minimized by applying a previously trained model. This could be extremely useful in ubiquitous environments that are characterized by the existence of resource constrained devices. To deal with the aforementioned scenario, meta-models can be used in the process of enhancing the drift detection mechanisms used by data stream algorithms, by representing and predicting when the change will occur. There are some real-world situations where a concept reappears, as in the case of intrusion detection systems (IDS), where the same incidents or an adaptation of them usually reappear over time. In these environments the early prediction of drift by means of a better knowledge of past models can help to anticipate to the change, thus improving efficiency of the model regarding the training instances needed. By means of using meta-models as a recurrent drift detection mechanism, the ability to share concepts representations among different data mining processes is open. That kind of exchanges could improve the accuracy of the resultant local model as such model may benefit from patterns similar to the local concept that were observed in other scenarios, but not yet locally. This would also improve the efficiency of training instances used during the classification process, as long as the exchange of models would aid in the application of already trained recurrent models, that have been previously seen by any of the collaborative devices. Which it is to say that the scope of recurrence detection and representation is broaden. In fact the detection, representation and exchange of concept drift patterns would be extremely useful for the law enforcement activities fighting against cyber crime. Being the information exchange one of the main pillars of cooperation, national units would benefit from the experience and knowledge gained by third parties. Moreover, in the specific scope of critical infrastructures protection it is crucial to count with information exchange mechanisms, both from a strategical and technical scope. The exchange of concept drift detection schemes in cyber security environments would aid in the process of preventing, detecting and effectively responding to threads in cyber space. Furthermore, as a complement of meta-models, a mechanism to assess the similarity between classification models is also needed when dealing with recurrent concepts. In this context, when reusing a previously trained model a rough comparison between concepts is usually made, applying boolean logic. The introduction of fuzzy logic comparisons between models could lead to a better efficient reuse of previously seen concepts, by applying not just equal models, but also similar ones. This work faces the aforementioned open issues by means of: the MMPRec system, that integrates a meta-model mechanism and a fuzzy similarity function; a collaborative environment to share meta-models between different devices; a recurrent drift generator that allows to test the usefulness of recurrent drift systems, as it is the case of MMPRec. Moreover, this thesis presents an experimental validation of the proposed contributions using synthetic and real datasets.