Semi-supervised subspace clustering and applications to neuroscience

Machine learning techniques are used for extracting valuable knowledge from data. Nowa¬days, these techniques are becoming even more important due to the evolution in data ac¬quisition and storage, which is leading to data with different characteristics that must be exploited. Therefore, advances in data collection must be accompanied with advances in machine learning techniques to solve new challenges that might arise, on both academic and real applications. There are several machine learning techniques depending on both data characteristics and purpose. Unsupervised classification or clustering is one of the most known techniques when data lack of supervision (unlabeled data) and the aim is to discover data groups (clusters) according to their similarity. On the other hand, supervised classification needs data with supervision (labeled data) and its aim is to make predictions about labels of new data. The presence of data labels is a very important characteristic that guides not only the learning task but also other related tasks such as validation. When only some of the available data are labeled whereas the others remain unlabeled (partially labeled data), neither clustering nor supervised classification can be used. This scenario, which is becoming common nowadays because of labeling process ignorance or cost, is tackled with semi-supervised learning techniques. This thesis focuses on the branch of semi-supervised learning closest to clustering, i.e., to discover clusters using available labels as support to guide and improve the clustering process. Another important data characteristic, different from the presence of data labels, is the relevance or not of data features. Data are characterized by features, but it is possible that not all of them are relevant, or equally relevant, for the learning process. A recent clustering tendency, related to data relevance and called subspace clustering, claims that different clusters might be described by different feature subsets. This differs from traditional solutions to data relevance problem, where a single feature subset (usually the complete set of original features) is found and used to perform the clustering process. The proximity of this work to clustering leads to the first goal of this thesis. As commented above, clustering validation is a difficult task due to the absence of data labels. Although there are many indices that can be used to assess the quality of clustering solutions, these validations depend on clustering algorithms and data characteristics. Hence, in the first goal three known clustering algorithms are used to cluster data with outliers and noise, to critically study how some of the most known validation indices behave. The main goal of this work is however to combine semi-supervised clustering with subspace clustering to obtain clustering solutions that can be correctly validated by using either known indices or expert opinions. Two different algorithms are proposed from different points of view to discover clusters characterized by different subspaces. For the first algorithm, available data labels are used for searching for subspaces firstly, before searching for clusters. This algorithm assigns each instance to only one cluster (hard clustering) and is based on mapping known labels to subspaces using supervised classification techniques. Subspaces are then used to find clusters using traditional clustering techniques. The second algorithm uses available data labels to search for subspaces and clusters at the same time in an iterative process. This algorithm assigns each instance to each cluster based on a membership probability (soft clustering) and is based on integrating known labels and the search for subspaces into a model-based clustering approach. The different proposals are tested using different real and synthetic databases, and comparisons to other methods are also included when appropriate. Finally, as an example of real and current application, different machine learning tech¬niques, including one of the proposals of this work (the most sophisticated one) are applied to a task of one of the most challenging biological problems nowadays, the human brain model¬ing. Specifically, expert neuroscientists do not agree with a neuron classification for the brain cortex, which makes impossible not only any modeling attempt but also the day-to-day work without a common way to name neurons. Therefore, machine learning techniques may help to get an accepted solution to this problem, which can be an important milestone for future research in neuroscience. Resumen Las técnicas de aprendizaje automático se usan para extraer información valiosa de datos. Hoy en día, la importancia de estas técnicas está siendo incluso mayor, debido a que la evolución en la adquisición y almacenamiento de datos está llevando a datos con diferentes características que deben ser explotadas. Por lo tanto, los avances en la recolección de datos deben ir ligados a avances en las técnicas de aprendizaje automático para resolver nuevos retos que pueden aparecer, tanto en aplicaciones académicas como reales. Existen varias técnicas de aprendizaje automático dependiendo de las características de los datos y del propósito. La clasificación no supervisada o clustering es una de las técnicas más conocidas cuando los datos carecen de supervisión (datos sin etiqueta), siendo el objetivo descubrir nuevos grupos (agrupaciones) dependiendo de la similitud de los datos. Por otra parte, la clasificación supervisada necesita datos con supervisión (datos etiquetados) y su objetivo es realizar predicciones sobre las etiquetas de nuevos datos. La presencia de las etiquetas es una característica muy importante que guía no solo el aprendizaje sino también otras tareas relacionadas como la validación. Cuando solo algunos de los datos disponibles están etiquetados, mientras que el resto permanece sin etiqueta (datos parcialmente etiquetados), ni el clustering ni la clasificación supervisada se pueden utilizar. Este escenario, que está llegando a ser común hoy en día debido a la ignorancia o el coste del proceso de etiquetado, es abordado utilizando técnicas de aprendizaje semi-supervisadas. Esta tesis trata la rama del aprendizaje semi-supervisado más cercana al clustering, es decir, descubrir agrupaciones utilizando las etiquetas disponibles como apoyo para guiar y mejorar el proceso de clustering. Otra característica importante de los datos, distinta de la presencia de etiquetas, es la relevancia o no de los atributos de los datos. Los datos se caracterizan por atributos, pero es posible que no todos ellos sean relevantes, o igualmente relevantes, para el proceso de aprendizaje. Una tendencia reciente en clustering, relacionada con la relevancia de los datos y llamada clustering en subespacios, afirma que agrupaciones diferentes pueden estar descritas por subconjuntos de atributos diferentes. Esto difiere de las soluciones tradicionales para el problema de la relevancia de los datos, en las que se busca un único subconjunto de atributos (normalmente el conjunto original de atributos) y se utiliza para realizar el proceso de clustering. La cercanía de este trabajo con el clustering lleva al primer objetivo de la tesis. Como se ha comentado previamente, la validación en clustering es una tarea difícil debido a la ausencia de etiquetas. Aunque existen muchos índices que pueden usarse para evaluar la calidad de las soluciones de clustering, estas validaciones dependen de los algoritmos de clustering utilizados y de las características de los datos. Por lo tanto, en el primer objetivo tres conocidos algoritmos se usan para agrupar datos con valores atípicos y ruido para estudiar de forma crítica cómo se comportan algunos de los índices de validación más conocidos. El objetivo principal de este trabajo sin embargo es combinar clustering semi-supervisado con clustering en subespacios para obtener soluciones de clustering que puedan ser validadas de forma correcta utilizando índices conocidos u opiniones expertas. Se proponen dos algoritmos desde dos puntos de vista diferentes para descubrir agrupaciones caracterizadas por diferentes subespacios. Para el primer algoritmo, las etiquetas disponibles se usan para bus¬car en primer lugar los subespacios antes de buscar las agrupaciones. Este algoritmo asigna cada instancia a un único cluster (hard clustering) y se basa en mapear las etiquetas cono-cidas a subespacios utilizando técnicas de clasificación supervisada. El segundo algoritmo utiliza las etiquetas disponibles para buscar de forma simultánea los subespacios y las agru¬paciones en un proceso iterativo. Este algoritmo asigna cada instancia a cada cluster con una probabilidad de pertenencia (soft clustering) y se basa en integrar las etiquetas conocidas y la búsqueda en subespacios dentro de clustering basado en modelos. Las propuestas son probadas utilizando diferentes bases de datos reales y sintéticas, incluyendo comparaciones con otros métodos cuando resulten apropiadas. Finalmente, a modo de ejemplo de una aplicación real y actual, se aplican diferentes técnicas de aprendizaje automático, incluyendo una de las propuestas de este trabajo (la más sofisticada) a una tarea de uno de los problemas biológicos más desafiantes hoy en día, el modelado del cerebro humano. Específicamente, expertos neurocientíficos no se ponen de acuerdo en una clasificación de neuronas para la corteza cerebral, lo que imposibilita no sólo cualquier intento de modelado sino también el trabajo del día a día al no tener una forma estándar de llamar a las neuronas. Por lo tanto, las técnicas de aprendizaje automático pueden ayudar a conseguir una solución aceptada para este problema, lo cual puede ser un importante hito para investigaciones futuras en neurociencia.

Assessing the dynamic behavior of wsn motes and rfid semi-passive tags for temperature monitoring.

A notorious advantage of wireless transmission is a significant reduction and simplification in wiring and harness. There are a lot of applications of wireless systems, but in many occasions sensor nodes require a specific housing to protect the electronics from hush environmental conditions. Nowadays the information is scarce and nonspecific on the dynamic behaviour of WSN and RFID. Therefore the purpose of this study is to evaluate the dynamic behaviour of the sensors. A series of trials were designed and performed covering temperature steps between cold room (5 °C), room temperature (23 °C) and heated environment (35 °C). As sensor nodes: three Crossbow motes, a surface mounted Nlaza module (with sensor Sensirion located on the motherboard), an aerial mounted Nlaza where the Sensirion sensor stayed at the end of a cable), and four tags RFID Turbo Tag (T700 model with and without housing), and 702-B (with and without housing). To assess the dynamic behaviour a first order response approach is used and fitted with dedicated optimization tools programmed in Matlab that allow extracting the time response (?) and corresponding determination coefficient (r2) with regard to experimental data. The shorter response time (20.9 s) is found for the uncoated T 700 tag which encapsulated version provides a significantly higher response (107.2 s). The highest ? corresponds to the Crossbow modules (144.4 s), followed by the surface mounted Nlaza module (288.1 s), while the module with aerial mounted sensor gives a response certainly close above to the T700 without coating (42.8 s). As a conclusion, the dynamic response of temperature sensors within wireless and RFID nodes is dramatically influenced by the way they are housed (to protect them from the environment) as well as by the heat released by the node electronics itself; its characterization is basic to allow monitoring of high rate temperature changes and to certify the cold chain. Besides the time to rise and to recover is significantly different being mostly higher for the latter than for the former.

Integrated assessment of policy interventions for promoting sustainable irrigation in semi-arid environments: A hydro-economic modeling approach

Sustaining irrigated agriculture to meet food production needs while maintaining aquatic ecosystems is at the heart of many policy debates in various parts of the world, especially in arid and semi-arid areas. Researchers and practitioners are increasingly calling for integrated approaches, and policy-makers are progressively supporting the inclusion of ecological and social aspects in water management programs. This paper contributes to this policy debate by providing an integrated economic-hydrologic modeling framework that captures the socio-economic and environmental effects of various policy initiatives and climate variability. This modeling integration includes a risk-based economic optimization model and a hydrologic water management simulation model that have been specified for the Middle Guadiana basin, a vulnerable drought-prone agro-ecological area with highly regulated river systems in southwest Spain. Namely, two key water policy interventions were investigated: the implementation of minimum environmental flows (supported by the European Water Framework Directive, EU WFD), and a reduction in the legal amount of water delivered for irrigation (planned measure included in the new Guadiana River Basin Management Plan, GRBMP, still under discussion). Results indicate that current patterns of excessive water use for irrigation in the basin may put environmental flow demands at risk, jeopardizing the WFD s goal of restoring the ?good ecological status? of water bodies by 2015. Conflicts between environmental and agricultural water uses will be stressed during prolonged dry episodes, and particularly in summer low-flow periods, when there is an important increase of crop irrigation water requirements. Securing minimum stream flows would entail a substantial reduction in irrigation water use for rice cultivation, which might affect the profitability and economic viability of small rice-growing farms located upstream in the river. The new GRBMP could contribute to balance competing water demands in the basin and to increase economic water productivity, but might not be sufficient to ensure the provision of environmental flows as required by the WFD. A thoroughly revision of the basin s water use concession system for irrigation seems to be needed in order to bring the GRBMP in line with the WFD objectives. Furthermore, the study illustrates that social, economic, institutional, and technological factors, in addition to bio-physical conditions, are important issues to be considered for designing and developing water management strategies. The research initiative presented in this paper demonstrates that hydro-economic models can explicitly integrate all these issues, constituting a valuable tool that could assist policy makers for implementing sustainable irrigation policies.

Carbon dioxide emissions from semi-arid soils amended with biochar alone or combined with mineral and organic fertilizers

Semi-arid soils cover a significant area of Earth s land surface and typically contain large amounts of inorganic C. Determining the effects of biochar additions on CO2 emissions fromsemi-arid soils is therefore essential for evaluating the potential of biochar as a climate change mitigation strategy. Here, we measured the CO2 that evolved from semi-arid calcareous soils amended with biochar at rates of 0 and 20 t ha?1 in a full factorial combination with three different fertilizers (mineral fertilizer, municipal solid waste compost, and sewage sludge) applied at four rates (equivalent to 0, 75, 150, and 225 kg potentially available N ha?1) during 182 days of aerobic incubation. A double exponential model, which describes cumulative CO2 emissions from two active soil C compartments with different turnover rates (one relatively stable and the other more labile), was found to fit verywell all the experimental datasets. In general, the organic fertilizers increased the size and decomposition rate of the stable and labile soil C pools. In contrast, biochar addition had no effects on any of the double exponential model parameters and did not interact with the effects ascribed to the type and rate of fertilizer. After 182 days of incubation, soil organic and microbial biomass C contents tended to increase with increasing the application rates of organic fertilizer, especially of compost, whereas increasing the rate of mineral fertilizer tended to suppress microbial biomass. Biochar was found to increase both organic and inorganic C contents in soil and not to interactwith the effects of type and rate of fertilizer on C fractions. As a whole, our results suggest that the use of biochar as enhancer of semi-arid soils, either alone or combined with mineral and organic fertilizers, is unlikely to increase abiotic and biotic soil CO2 emissions.

Design of a Semi-analytical Method to Describe Plasma-Wall Interactions

Electric probes are objects immersed in the plasma with sharp boundaries which collect of emit charged particles. Consequently, the nearby plasma evolves under abrupt imposed and/or naturally emerging conditions. There could be localized currents, different time scales for plasma species evolution, charge separation and absorbing-emitting walls. The traditional numerical schemes based on differences often transform these disparate boundary conditions into computational singularities. This is the case of models using advection-diffusion differential equations with source-sink terms (also called Fokker-Planck equations). These equations are used in both, fluid and kinetic descriptions, to obtain the distribution functions or the density for each plasma species close to the boundaries. We present a resolution method grounded on an integral advancing scheme by using approximate Green's functions, also called short-time propagators. All the integrals, as a path integration process, are numerically calculated, what states a robust grid-free computational integral method, which is unconditionally stable for any time step. Hence, the sharp boundary conditions, as the current emission from a wall, can be treated during the short-time regime providing solutions that works as if they were known for each time step analytically. The form of the propagator (typically a multivariate Gaussian) is not unique and it can be adjusted during the advancing scheme to preserve the conserved quantities of the problem. The effects of the electric or magnetic fields can be incorporated into the iterative algorithm. The method allows smooth transitions of the evolving solutions even when abrupt discontinuities are present. In this work it is proposed a procedure to incorporate, for the very first time, the boundary conditions in the numerical integral scheme. This numerical scheme is applied to model the plasma bulk interaction with a charge-emitting electrode, dealing with fluid diffusion equations combined with Poisson equation self-consistently. It has been checked the stability of this computational method under any number of iterations, even for advancing in time electrons and ions having different time scales. This work establishes the basis to deal in future work with problems related to plasma thrusters or emissive probes in electromagnetic fields.

Aerodynamic external pressure loads on a semi-circular bluff body under wind gusts

This brief communication concerns the unsteady aerodynamic external pressure loads acting on a semi-circular bluff body lying on a floor under wind gusts and describes the theoretical model, experimental setup, and experimental results obtained. The experimental setup is based on an open circuit, closed test section, low speed wind tunnel, which includes a sinusoidal gust generating mechanism, designed and built at the Instituto de Microgravedad “Ignacio Da Riva” of the Universidad Politécnica de Madrid (IDR/UPM). Based on the potential flow theory, a theoretical model has been proposed to analyse the problem, and experimental tests have been performed to study the unsteady aerodynamic loads on a semi-circular bluff body. By fitting the theoretical model predictions with the experimental results, influencing parameters of the unsteady aerodynamic loads are ascertained. The values of these parameters can help in clarifying the phenomenon of the external pressure loads on semi-circular bluff body under various gust frequencies. The theoretical model proposed allows the pressure variation to be split into two contributions, a quasi-steady term and an unsteady term with a simple physical meaning

The folding mechanism of larger model proteins: role of native structure.

The folding mechanism of a 125-bead heteropolymer model for proteins is investigated with Monte Carlo simulations on a cubic lattice. Sequences that do and do not fold in a reasonable time are compared. The overall folding behavior is found to be more complex than that of models for smaller proteins. Folding begins with a rapid collapse followed by a slow search through the semi-compact globule for a sequence-dependent stable core with about 30 out of 176 native contacts which serves as the transition state for folding to a near-native structure. Efficient search for the core is dependent on structural features of the native state. Sequences that fold have large amounts of stable, cooperative structure that is accessible through short-range initiation sites, such as those in anti-parallel sheets connected by turns. Before folding is completed, the system can encounter a second bottleneck, involving the condensation and rearrangement of surface residues. Overly stable local structure of the surface residues slows this stage of the folding process. The relation of the results from the 125-mer model studies to the folding of real proteins is discussed.

Atenuação de vibrações em pás de helicópteros utilizando circuito piezelétrico semi-passivo

O uso de materiais inteligentes em problemas de controle de vibração tem sido investigado em diversas pesquisas ao longo dos últimos anos. Apesar de que diferentes materiais inteligentes estão disponíveis, o piezelétrico tem recebido grande atenção devido à facilidade de uso como sensores, atuadores, ou ambos simultaneamente. As principais técnicas de controle usando materiais piezoelétricos são os ativos e passivos. Circuitos piezelétricos passivos são ajustados para uma frequência específica e, portanto, a largura de banda efetiva é pequena. Embora os sistemas ativos possam apresentar um bom desempenho no controle de vibração, a quantidade de energia externa e hardware adicionado são questões importantes. As técnicas SSD (Synchronized Switch Damping) foram desenvolvidas como uma alternativa aos controladores passivos e controladores ativos de vibração. Elas podem ser técnicas semi-ativas ou semi-passivas que introduzem um tratamento não linear na tensão elétrica proveniente do material piezelétrico e induz um aumento na conversão de energia mecânica para energia elétrica e, consequentemente, um aumento no efeito de amortecimento. Neste trabalho, o controle piezoelétrico semi-passivo de uma pá piezelétrica engastada é apresentado e comparado com outros controladores. O modelo não linear electromecânico de uma pá com piezocerâmicas incorporados é determinado com base no método variacional-assintótico (VAM). O sistema rotativo acoplado não linear é resolvido no domínio do tempo, utilizando um método de integração alfa-generalizado afim de garantir a estabilidade numérica. As simulações são realizadas para uma vasta gama de velocidades de rotação. Em primeiro lugar, um conjunto de resistências (variando desde a condição de curto-circuito para a condição de circuito aberto) é considerada. O efeito da resistência ótima (que resulta em máximo amortecimento) sobre o comportamento do sistema é investigado para o aumento da velocidade de rotação. Mais tarde, a técnica SSDS é utilizada para amortecer as oscilações da pá com o aumento da velocidade de rotação. Os resultados mostram que a técnica SSDS pode ser um método útil para o controle de vibrações de vigas rotativas não lineares, tais como pás de helicóptero.

Exact and analytic-numerical solutions of lagging models of heat transfer in a semi-infinite medium

Different non-Fourier models of heat conduction have been considered in recent years, in a growing area of applications, to model microscale and ultrafast, transient, nonequilibrium responses in heat and mass transfer. In this work, using Fourier transforms, we obtain exact solutions for different lagging models of heat conduction in a semi-infinite domain, which allow the construction of analytic-numerical solutions with prescribed accuracy. Examples of numerical computations, comparing the properties of the models considered, are presented.

Robust Solutions of MultiObjective Linear Semi-Infinite Programs under Constraint Data Uncertainty

The multiobjective optimization model studied in this paper deals with simultaneous minimization of finitely many linear functions subject to an arbitrary number of uncertain linear constraints. We first provide a radius of robust feasibility guaranteeing the feasibility of the robust counterpart under affine data parametrization. We then establish dual characterizations of robust solutions of our model that are immunized against data uncertainty by way of characterizing corresponding solutions of robust counterpart of the model. Consequently, we present robust duality theorems relating the value of the robust model with the corresponding value of its dual problem.

Controlo fuzzy semi-activo para aparelhos de amortecimento variável na protecção sísmica

Trabalho apresentado no 10º Congresso Nacional de Sismologia e Engenharia Sísmica, 20-22 abril de 2016, Ponta Delgada, Açores, Portugal

Graph-based Semi-Supervised Learning in Acoustic Modeling for Automatic Speech Recognition

Thesis (Ph.D.)--University of Washington, 2016-06

Using DEM to model ore breakage within a pilot scale SAG mill

The best accepted method for design of autogenous and semi-autogenous (AG/SAG) mills is to carry out pilot scale test work using a 1.8 m diameter by 0.6 m long pilot scale test mill. The load in such a mill typically contains 250,000-450,000 particles larger than 6 mm, allowing correct representation of more than 90% of the charge in Discrete Element Method (DEM) simulations. Most AG/SAG mills use discharge grate slots which are 15 mm or more in width. The mass in each size fraction usually decreases rapidly below grate size. This scale of DEM model is now within the possible range of standard workstations running an efficient DEM code. This paper describes various ways of extracting collision data front the DEM model and translating it into breakage estimates. Account is taken of the different breakage mechanisms (impact and abrasion) and of the specific impact histories of the particles in order to assess the breakage rates for various size fractions in the mills. At some future time, the integration of smoothed particle hydrodynamics with DEM will allow for the inclusion of slurry within the pilot mill simulation. (C) 2004 Elsevier Ltd. All rights reserved.

NMR imaging of the diffusion of water at 310 K into semi-IPNs of PEM and Poly(HEMA-co-THFMA)

Magnetic resonance imaging has been used to monitor the diffusion of water at 310 K into a series of semi-IPNs of poly(ethyl methacrylate), PEM, and copolymers of 2-hydroxyethyl methacrylate, HEMA, and tetrahydrofurfuryl methacrylate, THFMA. The diffusion was found to be well described by a Fickian kinetic model in the early stages of the water sorption process, and the diffusion coefficients were found to be slightly smaller than those for the copolymers of HEMA and THFMA, P(HEMA-co-THFMA), containing the same mole fraction of HEMA in the matrix. A second stage sorption process was identified in the later stage of water sorption by the PEM/PTHFMA semi-IPN and for the systems containing a P(HEMA-co-THFMA) component with a mole fraction HEMA of 0.6 or less. This was characterized by the presence of Water near the surface of the cylinders with a longer NMR T-2 relaxation time, which would be characteristic of mobile water, such as water present in large pores or surface fissures. The presence of the drug chlorhexidine in the polymer matrixes at a concentration of 5.625 wt % was found not to modify the properties significantly, but the diffusion coefficients for the water sorption were systematically smaller when the drug was present.

The displacement and strain field of three-dimensional rheologic model of earthquake preparation

Based on the three-dimensional elastic inclusion model proposed by Dobrovolskii, we developed a rheological inclusion model to study earthquake preparation processes. By using the Corresponding Principle in the theory of rheologic mechanics, we derived the analytic expressions of viscoelastic displacement U(r, t) , V(r, t) and W(r, t), normal strains epsilon(xx) (r, t), epsilon(yy) (r, t) and epsilon(zz) (r, t) and the bulk strain theta (r, t) at an arbitrary point (x, y, z) in three directions of X axis, Y axis and Z axis produced by a three-dimensional inclusion in the semi-infinite rheologic medium defined by the standard linear rheologic model. Subsequent to the spatial-temporal variation of bulk strain being computed on the ground produced by such a spherical rheologic inclusion, interesting results are obtained, suggesting that the bulk strain produced by a hard inclusion change with time according to three stages (alpha, beta, gamma) with different characteristics, similar to that of geodetic deformation observations, but different with the results of a soft inclusion. These theoretical results can be used to explain the characteristics of spatial-temporal evolution, patterns, quadrant-distribution of earthquake precursors, the changeability, spontaneity and complexity of short-term and imminent-term precursors. It offers a theoretical base to build physical models for earthquake precursors and to predict the earthquakes.