Semi-supervised classification based on subspace sparse representation

Graph plays an important role in graph-based semi-supervised classification. However, due to noisy and redundant features in high-dimensional data, it is not a trivial job to construct a well-structured graph on high-dimensional samples. In this paper, we take advantage of sparse representation in random subspaces for graph construction and propose a method called Semi-Supervised Classification based on Subspace Sparse Representation, SSC-SSR in short. SSC-SSR first generates several random subspaces from the original space and then seeks sparse representation coefficients in these subspaces. Next, it trains semi-supervised linear classifiers on graphs that are constructed by these coefficients. Finally, it combines these classifiers into an ensemble classifier by minimizing a linear regression problem. Unlike traditional graph-based semi-supervised classification methods, the graphs of SSC-SSR are data-driven instead of man-made in advance. Empirical study on face images classification tasks demonstrates that SSC-SSR not only has superior recognition performance with respect to competitive methods, but also has wide ranges of effective input parameters.

Prediction interval-based modelling of polymerization reactor: a new modelling strategy for chemical reactors

Precise and reliable modelling of polymerization reactor is challenging due to its complex reaction mechanism and non-linear nature. Researchers often make several assumptions when deriving theories and developing models for polymerization reactor. Therefore, traditional available models suffer from high prediction error. In contrast, data-driven modelling techniques provide a powerful framework to describe the dynamic behaviour of polymerization reactor. However, the traditional NN prediction performance is significantly dropped in the presence of polymerization process disturbances. Besides, uncertainty effects caused by disturbances present in reactor operation can be properly quantified through construction of prediction intervals (PIs) for model outputs. In this study, we propose and apply a PI-based neural network (PI-NN) model for the free radical polymerization system. This strategy avoids assumptions made in traditional modelling techniques for polymerization reactor system. Lower upper bound estimation (LUBE) method is used to develop PI-NN model for uncertainty quantification. To further improve the quality of model, a new method is proposed for aggregation of upper and lower bounds of PIs obtained from individual PI-NN models. Simulation results reveal that combined PI-NN performance is superior to those individual PI-NN models in terms of PI quality. Besides, constructed PIs are able to properly quantify effects of uncertainties in reactor operation, where these can be later used as part of the control process. © 2014 Taiwan Institute of Chemical Engineers.

Bootstrapping mobile app development

Modern IDEs provide limited support for developers when starting a new data-driven mobile app. App developers are currently required to write copious amounts of boilerplate code, scripts, organise complex directories, and author actual functionality. Although this scenario is ripe for automation, current tools are yet to address it adequately. In this paper we present RAPPT, a tool that generates the scaffolding of a mobile app based on a high level description specified in a Domain Specific Language (DSL). We demonstrate the feasibility of our approach by an example case study and feedback from a professional development team. Demo at: https://www.youtube.com/watch?v=ffquVgBYpLM.

Approximation of centroid end-points and switch points for replacing type reduction algorithms

Despite several years of research, type reduction (TR) operation in interval type-2 fuzzy logic system (IT2FLS) cannot perform as fast as a type-1 defuzzifier. In particular, widely used Karnik-Mendel (KM) TR algorithm is computationally much more demanding than alternative TR approaches. In this work, a data driven framework is proposed to quickly, yet accurately, estimate the output of the KM TR algorithm using simple regression models. Comprehensive simulation performed in this study shows that the centroid end-points of KM algorithm can be approximated with a mean absolute percentage error as low as 0.4%. Also, switch point prediction accuracy can be as high as 100%. In conjunction with the fact that simple regression model can be trained with data generated using exhaustive defuzzification method, this work shows the potential of proposed method to provide highly accurate, yet extremely fast, TR approximation method. Speed of the proposed method should theoretically outperform all available TR methods while keeping the uncertainty information intact in the process.

Linear approximation of Karnik-Mendel type reduction algorithm

Karnik-Mendel (KM) algorithm is the most used and researched type reduction (TR) algorithm in literature. This algorithm is iterative in nature and despite consistent long term effort, no general closed form formula has been found to replace this computationally expensive algorithm. In this research work, we demonstrate that the outcome of KM algorithm can be approximated by simple linear regression techniques. Since most of the applications will have a fixed range of inputs with small scale variations, it is possible to handle those complexities in design phase and build a fuzzy logic system (FLS) with low run time computational burden. This objective can be well served by the application of regression techniques. This work presents an overview of feasibility of regression techniques for design of data-driven type reducers while keeping the uncertainty bound in FLS intact. Simulation results demonstrates the approximation error is less than 2%. Thus our work preserve the essence of Karnik-Mendel algorithm and serves the requirement of low
computational complexities.

Switch point finding using polynomial regression for fuzzy type reduction algorithms

Karnik-Mendel (KM) algorithm is the most widely used type reduction (TR) method in literature for the design of interval type-2 fuzzy logic systems (IT2FLS). Its iterative nature for finding left and right switch points is its Achilles heel. Despite a decade of research, none of the alternative TR methods offer uncertainty measures equivalent to KM algorithm. This paper takes a data-driven approach to tackle the computational burden of this algorithm while keeping its key features. We propose a regression method to approximate left and right switch points found by KM algorithm. Approximator only uses the firing intervals, rnles centroids, and FLS strnctural features as inputs. Once training is done, it can precisely approximate the left and right switch points through basic vector multiplications. Comprehensive simulation results demonstrate that the approximation accuracy for a wide variety of FLSs is 100%. Flexibility, ease of implementation, and speed are other features of the proposed method.

Effect Of Different Hydrological Model Structures On The Assimilation Of Distributed Uncertain Observations

The reliable evaluation of the flood forecasting is a crucial problem for assessing flood risk and consequent damages. Different hydrological models (distributed, semi-distributed or lumped) have been proposed in order to deal with this issue. The choice of the proper model structure has been investigated by many authors and it is one of the main sources of uncertainty for a correct evaluation of the outflow hydrograph. In addition, the recent increasing of data availability makes possible to update hydrological models as response of real-time observations. For these reasons, the aim of this work it is to evaluate the effect of different structure of a semi-distributed hydrological model in the assimilation of distributed uncertain discharge observations. The study was applied to the Bacchiglione catchment, located in Italy. The first methodological step was to divide the basin in different sub-basins according to topographic characteristics. Secondly, two different structures of the semi-distributed hydrological model were implemented in order to estimate the outflow hydrograph. Then, synthetic observations of uncertain value of discharge were generated, as a function of the observed and simulated value of flow at the basin outlet, and assimilated in the semi-distributed models using a Kalman Filter. Finally, different spatial patterns of sensors location were assumed to update the model state as response of the uncertain discharge observations. The results of this work pointed out that, overall, the assimilation of uncertain observations can improve the hydrologic model performance. In particular, it was found that the model structure is an important factor, of difficult characterization, since can induce different forecasts in terms of outflow discharge. This study is partly supported by the FP7 EU Project WeSenseIt.

Prediction Of Hydrological Models’ Uncertainty By A Committee Of Machine Learning-Models

This study presents an approach to combine uncertainties of the hydrological model outputs predicted from a number of machine learning models. The machine learning based uncertainty prediction approach is very useful for estimation of hydrological models' uncertainty in particular hydro-metrological situation in real-time application [1]. In this approach the hydrological model realizations from Monte Carlo simulations are used to build different machine learning uncertainty models to predict uncertainty (quantiles of pdf) of the a deterministic output from hydrological model . Uncertainty models are trained using antecedent precipitation and streamflows as inputs. The trained models are then employed to predict the model output uncertainty which is specific for the new input data. We used three machine learning models namely artificial neural networks, model tree, locally weighted regression to predict output uncertainties. These three models produce similar verification results, which can be improved by merging their outputs dynamically. We propose an approach to form a committee of the three models to combine their outputs. The approach is applied to estimate uncertainty of streamflows simulation from a conceptual hydrological model in the Brue catchment in UK and the Bagmati catchment in Nepal. The verification results show that merged output is better than an individual model output. [1] D. L. Shrestha, N. Kayastha, and D. P. Solomatine, and R. Price. Encapsulation of parameteric uncertainty statistics by various predictive machine learning models: MLUE method, Journal of Hydroinformatic, in press, 2013.

Decision Tree Classification Model In Water Supply Network

With the service life of water supply network (WSN) growth, the growing phenomenon of aging pipe network has become exceedingly serious. As urban water supply network is hidden underground asset, it is difficult for monitoring staff to make a direct classification towards the faults of pipe network by means of the modern detecting technology. In this paper, based on the basic property data (e.g. diameter, material, pressure, distance to pump, distance to tank, load, etc.) of water supply network, decision tree algorithm (C4.5) has been carried out to classify the specific situation of water supply pipeline. Part of the historical data was used to establish a decision tree classification model, and the remaining historical data was used to validate this established model. Adopting statistical methods were used to access the decision tree model including basic statistical method, Receiver Operating Characteristic (ROC) and Recall-Precision Curves (RPC). These methods has been successfully used to assess the accuracy of this established classification model of water pipe network. The purpose of classification model was to classify the specific condition of water pipe network. It is important to maintain the pipeline according to the classification results including asset unserviceable (AU), near perfect condition (NPC) and serious deterioration (SD). Finally, this research focused on pipe classification which plays a significant role in maintaining water supply networks in the future.

Newly Automated System For Integrated Assessment Of The Conditions Of Underwater Gas Pipelines

An underwater gas pipeline is the portion of the pipeline that crosses a river beneath its bottom. Underwater gas pipelines are subject to increasing dangers as time goes by. An accident at an underwater gas pipeline can lead to technological and environmental disaster on the scale of an entire region. Therefore, timely troubleshooting of all underwater gas pipelines in order to prevent any potential accidents will remain a pressing task for the industry. The most important aspect of resolving this challenge is the quality of the automated system in question. Now the industry doesn't have any automated system that fully meets the needs of the experts working in the field maintaining underwater gas pipelines. Principle Aim of this Research: This work aims to develop a new system of automated monitoring which would simplify the process of evaluating the technical condition and decision making on planning and preventive maintenance and repair work on the underwater gas pipeline. Objectives: Creation a shared model for a new, automated system via IDEF3; Development of a new database system which would store all information about underwater gas pipelines; Development a new application that works with database servers, and provides an explanation of the results obtained from the server; Calculation of the values MTBF for specified pipelines based on quantitative data obtained from tests of this system. Conclusion: The new, automated system PodvodGazExpert has been developed for timely and qualitative determination of the physical conditions of underwater gas pipeline; The basis of the mathematical analysis of this new, automated system uses principal component analysis method; The process of determining the physical condition of an underwater gas pipeline with this new, automated system increases the MTBF by a factor of 8.18 above the existing system used today in the industry.

Leakage Detection And Localization Method Based On A Hybrid Inverse/Direct Modelling Approach Suitable For Handling Multiple-Leak Scenarios

When an accurate hydraulic network model is available, direct modeling techniques are very straightforward and reliable for on-line leakage detection and localization applied to large class of water distribution networks. In general, this type of techniques based on analytical models can be seen as an application of the well-known fault detection and isolation theory for complex industrial systems. Nonetheless, the assumption of single leak scenarios is usually made considering a certain leak size pattern which may not hold in real applications. Upgrading a leak detection and localization method based on a direct modeling approach to handle multiple-leak scenarios can be, on one hand, quite straightforward but, on the other hand, highly computational demanding for large class of water distribution networks given the huge number of potential water loss hotspots. This paper presents a leakage detection and localization method suitable for multiple-leak scenarios and large class of water distribution networks. This method can be seen as an upgrade of the above mentioned method based on a direct modeling approach in which a global search method based on genetic algorithms has been integrated in order to estimate those network water loss hotspots and the size of the leaks. This is an inverse / direct modeling method which tries to take benefit from both approaches: on one hand, the exploration capability of genetic algorithms to estimate network water loss hotspots and the size of the leaks and on the other hand, the straightforwardness and reliability offered by the availability of an accurate hydraulic model to assess those close network areas around the estimated hotspots. The application of the resulting method in a DMA of the Barcelona water distribution network is provided and discussed. The obtained results show that leakage detection and localization under multiple-leak scenarios may be performed efficiently following an easy procedure.

Integração de dados em ambiente SIG para mapeamento de favorabilidade mineral de ouro na ilha cristalina de Rivera (Uruguai)

Guias para exploração mineral são normalmente baseados em modelos conceituais de depósitos. Esses guias são, normalmente, baseados na experiência dos geólogos, em dados descritivos e em dados genéticos. Modelamentos numéricos, probabilísticos e não probabilísticos, para estimar a ocorrência de depósitos minerais é um novo procedimento que vem a cada dia aumentando sua utilização e aceitação pela comunidade geológica. Essa tese utiliza recentes metodologias para a geração de mapas de favorablidade mineral. A denominada Ilha Cristalina de Rivera, uma janela erosional da Bacia do Paraná, situada na porção norte do Uruguai, foi escolhida como estudo de caso para a aplicação das metodologias. A construção dos mapas de favorabilidade mineral foi feita com base nos seguintes tipos de dados, informações e resultados de prospecção: 1) imagens orbitais; 2) prospecção geoquimica; 3) prospecção aerogeofísica; 4) mapeamento geo-estrutural e 5) altimetria. Essas informacões foram selecionadas e processadas com base em um modelo de depósito mineral (modelo conceitual), desenvolvido com base na Mina de Ouro San Gregorio. O modelo conceitual (modelo San Gregorio), incluiu características descritivas e genéticas da Mina San Gregorio, a qual abrange os elementos característicos significativos das demais ocorrências minerais conhecidas na Ilha Cristalina de Rivera. A geração dos mapas de favorabilidade mineral envolveu a construção de um banco de dados, o processamento dos dados, e a integração dos dados. As etapas de construção e processamento dos dados, compreenderam a coleta, a seleção e o tratamento dos dados de maneira a constituírem os denominados Planos de Informação. Esses Planos de Informação foram gerados e processados organizadamente em agrupamentos, de modo a constituírem os Fatores de Integração para o mapeamento de favorabilidade mineral na Ilha Cristalina de Rivera. Os dados foram integrados por meio da utilização de duas diferentes metodologias: 1) Pesos de Evidência (dirigida pelos dados) e 2) Lógica Difusa (dirigida pelo conhecimento). Os mapas de favorabilidade mineral resultantes da implementação das duas metodologias de integração foram primeiramente analisados e interpretados de maneira individual. Após foi feita uma análise comparativa entre os resultados. As duas metodologias xxiv obtiveram sucesso em identificar, como áreas de alta favorabilidade, as áreas mineralizadas conhecidas, além de outras áreas ainda não trabalhadas. Os mapas de favorabilidade mineral resultantes das duas metodologias mostraram-se coincidentes em relação as áreas de mais alta favorabilidade. A metodologia Pesos de Evidência apresentou o mapa de favorabilidade mineral mais conservador em termos de extensão areal, porém mais otimista em termos de valores de favorabilidade em comparação aos mapas de favorabilidade mineral resultantes da implementação da metodologia Lógica Difusa. Novos alvos para exploração mineral foram identificados e deverão ser objeto de investigação em detalhe.

Estimação de regressões aditivas via backfitting e integração marginal: performance em pequenas amostras

Nesta dissertação realizou-se um experimento de Monte Carlo para re- velar algumas características das distribuições em amostras finitas dos estimadores Backfitting (B) e de Integração Marginal(MI) para uma regressão aditiva bivariada. Está-se particularmente interessado em fornecer alguma evidência de como os diferentes métodos de seleção da janela hn, tais co- mo os métodos plug-in, impactam as propriedades em pequenas amostras dos estimadores. Está-se interessado, também, em fornecer evidência do comportamento de diferentes estimadores de hn relativamente a seqüência ótima de hn que minimiza uma função perda escolhida. O impacto de ignorar a dependência entre os regressores na estimação da janela é tam- bém investigado. Esta é uma prática comum e deve ter impacto sobre o desempenho dos estimadores. Além disso, não há nenhuma rotina atual- mente disponível nos pacotes estatísticos/econométricos para a estimação de regressões aditivas via os métodos de Backfitting e Integração Marginal. É um dos objetivos a criação de rotinas em Gauss para a implementação prática destes estimadores. Por fim, diferentemente do que ocorre atual- mente, quando a utilização dos estimadores-B e MI é feita de maneira completamente ad-hoc, há o objetivo de fornecer a usuários informação que permita uma escolha mais objetiva de qual estimador usar quando se está trabalhando com uma amostra finita.