Model selection for discriminative restricted boltzmann machines through meta-heuristic techniques

Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq)

The use of artificial neural networks in analysing the nutritional ecology of Chrysomya megacephala (F.) (Diptera: Calliphoridae), compared with a statistical model

Artificial neural networks (ANNs) have been widely applied to the resolution of complex biological problems. An important feature of neural models is that their implementation is not precluded by the theoretical distribution shape of the data used. Frequently, the performance of ANNs over linear or non-linear regression-based statistical methods is deemed to be significantly superior if suitable sample sizes are provided, especially in multidimensional and non-linear processes. The current work was aimed at utilising three well-known neural network methods in order to evaluate whether these models would be able to provide more accurate outcomes in relation to a conventional regression method in pupal weight predictions of Chrysomya megacephala, a species of blowfly (Diptera: Calliphoridae), using larval density (i.e. the initial number of larvae), amount of available food and pupal size as input data. It was possible to notice that the neural networks yielded more accurate performances in comparison with the statistical model (multiple regression). Assessing the three types of networks utilised (Multi-layer Perceptron, Radial Basis Function and Generalised Regression Neural Network), no considerable differences between these models were detected. The superiority of these neural models over a classical statistical method represents an important fact, because more accurate models may clarify several intricate aspects concerning the nutritional ecology of blowflies.

Connections between the Sznajd model with general confidence rules and graph theory

The Sznajd model is a sociophysics model that is used to model opinion propagation and consensus formation in societies. Its main feature is that its rules favor bigger groups of agreeing people. In a previous work, we generalized the bounded confidence rule in order to model biases and prejudices in discrete opinion models. In that work, we applied this modification to the Sznajd model and presented some preliminary results. The present work extends what we did in that paper. We present results linking many of the properties of the mean-field fixed points, with only a few qualitative aspects of the confidence rule (the biases and prejudices modeled), finding an interesting connection with graph theory problems. More precisely, we link the existence of fixed points with the notion of strongly connected graphs and the stability of fixed points with the problem of finding the maximal independent sets of a graph. We state these results and present comparisons between the mean field and simulations in Barabasi-Albert networks, followed by the main mathematical ideas and appendices with the rigorous proofs of our claims and some graph theory concepts, together with examples. We also show that there is no qualitative difference in the mean-field results if we require that a group of size q > 2, instead of a pair, of agreeing agents be formed before they attempt to convince other sites (for the mean field, this would coincide with the q-voter model).

Hydration and anomalous solubility of the Bell-Lavis model as solvent

We address the investigation of the solvation properties of the minimal orientational model for water originally proposed by [Bell and Lavis, J. Phys. A 3, 568 (1970)]. The model presents two liquid phases separated by a critical line. The difference between the two phases is the presence of structure in the liquid of lower density, described through the orientational order of particles. We have considered the effect of a small concentration of inert solute on the solvent thermodynamic phases. Solute stabilizes the structure of solvent by the organization of solvent particles around solute particles at low temperatures. Thus, even at very high densities, the solution presents clusters of structured water particles surrounding solute inert particles, in a region in which pure solvent would be free of structure. Solute intercalates with solvent, a feature which has been suggested by experimental and atomistic simulation data. Examination of solute solubility has yielded a minimum in that property, which may be associated with the minimum found for noble gases. We have obtained a line of minimum solubility (TmS) across the phase diagram, accompanying the line of maximum density. This coincidence is easily explained for noninteracting solute and it is in agreement with earlier results in the literature. We give a simple argument which suggests that interacting solute would dislocate TmS to higher temperatures.

A heat transfer model of the human upper limbs

A steady state multi-segmented heat transfer model of the human upper limbs was developed. The main purpose was to evaluate the impact of blood flow through superficial veins and subcutaneous vascular structures in the palm of the hands over the heat transfer between the limbs and the environment. The distinguishing feature of the model is the inclusion of a detailed circulatory network composed of vessels with diameter larger than 1 mm. The model was validated by comparing its results from exposures to a hot, a neutral, and a cold environment to experimental data presented in the literature. (C) 2011 Elsevier Ltd. All rights reserved.

A feature selection approach for identification of signature genes from SAGE data

Abstract Background One goal of gene expression profiling is to identify signature genes that robustly distinguish different types or grades of tumors. Several tumor classifiers based on expression profiling have been proposed using microarray technique. Due to important differences in the probabilistic models of microarray and SAGE technologies, it is important to develop suitable techniques to select specific genes from SAGE measurements. Results A new framework to select specific genes that distinguish different biological states based on the analysis of SAGE data is proposed. The new framework applies the bolstered error for the identification of strong genes that separate the biological states in a feature space defined by the gene expression of a training set. Credibility intervals defined from a probabilistic model of SAGE measurements are used to identify the genes that distinguish the different states with more reliability among all gene groups selected by the strong genes method. A score taking into account the credibility and the bolstered error values in order to rank the groups of considered genes is proposed. Results obtained using SAGE data from gliomas are presented, thus corroborating the introduced methodology. Conclusion The model representing counting data, such as SAGE, provides additional statistical information that allows a more robust analysis. The additional statistical information provided by the probabilistic model is incorporated in the methodology described in the paper. The introduced method is suitable to identify signature genes that lead to a good separation of the biological states using SAGE and may be adapted for other counting methods such as Massive Parallel Signature Sequencing (MPSS) or the recent Sequencing-By-Synthesis (SBS) technique. Some of such genes identified by the proposed method may be useful to generate classifiers.

The characterization of the Atxn2-CAG42-knock-in mouse as a model for Spinocerebellar Ataxia Type 2

Die Ursache der neurodegenerativen Erkrankung Spinozerebelläre Ataxie Typ 2 (SCA2) ist eine expandierte Polyglutamin-Domäne im humanen ATXN2-Gen von normalerweise 22 auf über 31 CAGs. Von der Degeneration sind vorwiegend die zerebellären Purkinje Neuronen betroffen, in denen zunehmend zytoplasmatische Aggregate sichtbar werden. Auch wenn die genaue Funktion von ATXN2 und die zugrunde liegenden molekularen Mechanismen noch immer ungeklärt sind, werden ein toxischer Funktionsgewinn sowie der Verlust der normalen Proteinfunktion als mögliche Ursachen diskutiert.rnUm ein wirklichkeitsgetreues Tiermodell für die SCA2 zu haben, wurde eine knock-in Maus generiert, deren einzelnes CAG im Atxn2-Gen durch 42 CAGs ersetzt wurde. Dieses Mausmodell ist durch eine stabile Vererbung der Expansion charakterisiert. Weiterhin zeigt sie ein verringertes Körpergewicht sowie eine spät beginnende motorische Inkoordination, was dem Krankheitsbild von SCA2 entspricht. rnIm Weiteren konnte gezeigt werden, dass, obwohl die Atxn2 mRNA-Spiegel in Großhirn und Kleinhirn erhöht waren, die Menge an löslichem ATXN2 im Laufe der Zeit abnahm und dies mit einem Auftreten an unlöslichem ATXN2 korrelierte. Dieser im Kleinhirn progressive Prozess resultierte schließlich in zytoplasmatischen Aggregaten innerhalb der Purkinje Neuronen alter Mäuse. Der Verlust an löslichem ATXN2 könnte Effekte erklären, die auf einen partiellen Funktionsverlust von ATXN2 zurückzuführen sind, wobei die Aggregatbildung einen toxischen Funktionsgewinn wiederspiegeln könnte. Neben ATXN2 wurde auch sein Interaktor PABPC1 zunehmend unlöslich. Während dies im Großhirn eine Erhöhung der PABPC1 mRNA- und löslichen Proteinspiegel zur Folge hatte, konnte keine kompensatorische Veränderung seiner mRNA und zudem eine Verminderung an löslichem PABPC1 im Kleinhirn beobachtet werden. Auch PABPC1 wurde in Aggregate sequestriert. Diese Unterschiede zwischen Großhirn und Kleinhirn könnten zu der spezifischen Vulnerabilität des Kleinhirns beitragen.rnUm die Folgen auf mRNA-Prozessierung zu untersuchen, wurde ein Transkriptomprofil im mittleren sowie fortgeschrittenen Alter der Mäuse erstellt. Hierbei war eine erhöhte Expression von Fbxw8 im Kleinhirn alter Mäuse auffällig. Als Komponente eines Ubiquitin-E3-Ligase-Komplexes, hilft FBXW8 in der Degradierung von Zielproteinen und könnte somit die Toxizität des expandieren ATXN2 verringern. rnZur näheren Beschreibung der physiologischen Funktion von ATXN2, konnte in ATXN2-knock-out Mäusen gezeigt werden, dass das Fehlen von ATXN2 zu einer reduzierten globalen Proteinsyntheserate führte und somit eine Rolle als Translationsaktivator möglich erscheint. Kompensatorisch wurde eine erhöhte S6-Phosphorylierung gemessen.

Computer simulation methods to study interfacial tensions : from the Ising model to colloidal crystals

In condensed matter systems, the interfacial tension plays a central role for a multitude of phenomena. It is the driving force for nucleation processes, determines the shape and structure of crystalline structures and is important for industrial applications. Despite its importance, the interfacial tension is hard to determine in experiments and also in computer simulations. While for liquid-vapor interfacial tensions there exist sophisticated simulation methods to compute the interfacial tension, current methods for solid-liquid interfaces produce unsatisfactory results.rnrnAs a first approach to this topic, the influence of the interfacial tension on nuclei is studied within the three-dimensional Ising model. This model is well suited because despite its simplicity, one can learn much about nucleation of crystalline nuclei. Below the so-called roughening temperature, nuclei in the Ising model are not spherical anymore but become cubic because of the anisotropy of the interfacial tension. This is similar to crystalline nuclei, which are in general not spherical but more like a convex polyhedron with flat facets on the surface. In this context, the problem of distinguishing between the two bulk phases in the vicinity of the diffuse droplet surface is addressed. A new definition is found which correctly determines the volume of a droplet in a given configuration if compared to the volume predicted by simple macroscopic assumptions.rnrnTo compute the interfacial tension of solid-liquid interfaces, a new Monte Carlo method called ensemble switch method'' is presented which allows to compute the interfacial tension of liquid-vapor interfaces as well as solid-liquid interfaces with great accuracy. In the past, the dependence of the interfacial tension on the finite size and shape of the simulation box has often been neglected although there is a nontrivial dependence on the box dimensions. As a consequence, one needs to systematically increase the box size and extrapolate to infinite volume in order to accurately predict the interfacial tension. Therefore, a thorough finite-size scaling analysis is established in this thesis. Logarithmic corrections to the finite-size scaling are motivated and identified, which are of leading order and therefore must not be neglected. The astounding feature of these logarithmic corrections is that they do not depend at all on the model under consideration. Using the ensemble switch method, the validity of a finite-size scaling ansatz containing the aforementioned logarithmic corrections is carefully tested and confirmed. Combining the finite-size scaling theory with the ensemble switch method, the interfacial tension of several model systems, ranging from the Ising model to colloidal systems, is computed with great accuracy.

A triplet spike-timing–dependent plasticity model generalizes the Bienenstock–Cooper–Munro rule to higher-order spatiotemporal correlations

Synaptic strength depresses for low and potentiates for high activation of the postsynaptic neuron. This feature is a key property of the Bienenstock–Cooper–Munro (BCM) synaptic learning rule, which has been shown to maximize the selectivity of the postsynaptic neuron, and thereby offers a possible explanation for experience-dependent cortical plasticity such as orientation selectivity. However, the BCM framework is rate-based and a significant amount of recent work has shown that synaptic plasticity also depends on the precise timing of presynaptic and postsynaptic spikes. Here we consider a triplet model of spike-timing–dependent plasticity (STDP) that depends on the interactions of three precisely timed spikes. Triplet STDP has been shown to describe plasticity experiments that the classical STDP rule, based on pairs of spikes, has failed to capture. In the case of rate-based patterns, we show a tight correspondence between the triplet STDP rule and the BCM rule. We analytically demonstrate the selectivity property of the triplet STDP rule for orthogonal inputs and perform numerical simulations for nonorthogonal inputs. Moreover, in contrast to BCM, we show that triplet STDP can also induce selectivity for input patterns consisting of higher-order spatiotemporal correlations, which exist in natural stimuli and have been measured in the brain. We show that this sensitivity to higher-order correlations can be used to develop direction and speed selectivity.

UPLC-MS-based urine metabolomics reveals indole-3-lactic acid and phenyllactic acid as conserved biomarkers for alcohol-induced liver disease in the Ppara-null mouse model

Since the development and prognosis of alcohol-induced liver disease (ALD) vary significantly with genetic background, identification of a genetic background-independent noninvasive ALD biomarker would significantly improve screening and diagnosis. This study explored the effect of genetic background on the ALD-associated urinary metabolome using the Ppara-null mouse model on two different backgrounds, C57BL/6 (B6) and 129/SvJ (129S), along with their wild-type counterparts. Reversed-phase gradient UPLC-ESI-QTOF-MS analysis revealed that urinary excretion of a number of metabolites, such as ethylsulfate, 4-hydroxyphenylacetic acid, 4-hydroxyphenylacetic acid sulfate, adipic acid, pimelic acid, xanthurenic acid, and taurine, were background-dependent. Elevation of ethyl-β-d-glucuronide and N-acetylglycine was found to be a common signature of the metabolomic response to alcohol exposure in wild-type as well as in Ppara-null mice of both strains. However, increased excretion of indole-3-lactic acid and phenyllactic acid was found to be a conserved feature exclusively associated with the alcohol-treated Ppara-null mouse on both backgrounds that develop liver pathologies similar to the early stages of human ALD. These markers reflected the biochemical events associated with early stages of ALD pathogenesis. The results suggest that indole-3-lactic acid and phenyllactic acid are potential candidates for conserved and pathology-specific high-throughput noninvasive biomarkers for early stages of ALD.

A Computational Model for Processing Lexical Information"

The group analysed some syntactic and phonological phenomena that presuppose the existence of interrelated components within the lexicon, which motivate the assumption that there are some sublexicons within the global lexicon of a speaker. This result is confirmed by experimental findings in neurolinguistics. Hungarian speaking agrammatic aphasics were tested in several ways, the results showing that the sublexicon of closed-class lexical items provides a highly automated complex device for processing surface sentence structure. Analysing Hungarian ellipsis data from a semantic-syntactic aspect, the group established that the lexicon is best conceived of being as split into at least two main sublexicons: the store of semantic-syntactic feature bundles and a separate store of sound forms. On this basis they proposed a format for representing open-class lexical items whose meanings are connected via certain semantic relations. They also proposed a new classification of verbs to account for the contribution of the aspectual reading of the sentence depending on the referential type of the argument, and a new account of the syntactic and semantic behaviour of aspectual prefixes. The partitioned sets of lexical items are sublexicons on phonological grounds. These sublexicons differ in terms of phonotactic grammaticality. The degrees of phonotactic grammaticality are tied up with the problem of psychological reality, of how many degrees of this native speakers are sensitive to. The group developed a hierarchical construction network as an extension of the original General Inheritance Network formalism and this framework was then used as a platform for the implementation of the grammar fragments.

Model based experimental investigation on Powered Gait Orthosis (PGO)

Research on rehabilitation showed that appropriate and repetitive mechanical movements can help spinal cord injured individuals to restore their functional standing and walking. The objective of this paper was to achieve appropriate and repetitive joint movements and approximately normal gait through the PGO by replicating normal walking, and to minimize the energy consumption for both patients and the device. A model based experimental investigative approach is presented in this dissertation. First, a human model was created in Ideas and human walking was simulated in Adams. The main feature of this model was the foot ground contact model, which had distributed contact points along the foot and varied viscoelasticity. The model was validated by comparison of simulated results of normal walking and measured ones from the literature. It was used to simulate current PGO walking to investigate the real causes of poor function of the current PGO, even though it had joint movements close to normal walking. The direct cause was one leg moving at a time, which resulted in short step length and no clearance after toe off. It can not be solved by simply adding power on both hip joints. In order to find a better answer, a PGO mechanism model was used to investigate different walking mechanisms by locking or releasing some joints. A trade-off between energy consumption, control complexity and standing position was found. Finally a foot release PGO virtual model was created and simulated and only foot release mechanism was developed into a prototype. Both the release mechanism and the design of foot release were validated through the experiment by adding the foot release on the current PGO. This demonstrated an advancement in improving functional aspects of the current PGO even without a whole physical model of foot release PGO for comparison.

Origin of renal myofibroblasts in the model of unilateral ureter obstruction in the rat

Tubulo-interstitial fibrosis is a constant feature of chronic renal failure and it is suspected to contribute importantly to the deterioration of renal function. In the fibrotic kidney there exists, besides normal fibroblasts, a large population of myofibroblasts, which are supposedly responsible for the increased production of intercellular matrix. It has been proposed that myofibroblasts in chronic renal failure originate from the transformation of tubular cells via epithelial-mesenchymal transition (EMT) or from infiltration by bone marrow-derived precursors. Little attention has been paid to the possibility of a transformation of resident fibroblasts into myofibroblasts in renal fibrosis. Therefore we examined the fate of resident fibroblasts in the initial phase of renal fibrosis in the classical model of unilateral ureter obstruction (UUO) in the rat. Rats were perfusion-fixed on days 1, 2, 3 and 4 after ligature of the right ureter. Starting from 1 day of UUO an increasing expression of alpha-smooth muscle actin (alphaSMA) in resident fibroblasts was revealed by immunofluorescence and confirmed by the observation of bundles of microfilaments and webs of intermediate filaments in the electron microscope. Inversely, there was a decreased expression of 5'-nucleotidase (5'NT), a marker of renal cortical fibroblasts. The RER became more voluminous, suggesting an increased synthesis of matrix. Intercellular junctions, a characteristic feature of myofibroblasts, became more frequent. The mitotic activity in fibroblasts was strongly increased. Renal tubules underwent severe regressive changes but the cells retained their epithelial characteristics and there was no sign of EMT. In conclusion, after ureter ligature, resident peritubular fibroblasts proliferated and they showed progressive alterations, suggesting a transformation in myofibroblasts. Thus the resident fibroblasts likely play a central role in fibrosis in that model.

Structural model of the CopA copper ATPase of Enterococcus hirae based on chemical cross-linking

The CopA copper ATPase of Enterococcus hirae belongs to the family of heavy metal pumping CPx-type ATPases and shares 43% sequence similarity with the human Menkes and Wilson copper ATPases. Due to a lack of suitable protein crystals, only partial three-dimensional structures have so far been obtained for this family of ion pumps. We present a structural model of CopA derived by combining topological information obtained by intramolecular cross-linking with molecular modeling. Purified CopA was cross-linked with different bivalent reagents, followed by tryptic digestion and identification of cross-linked peptides by mass spectrometry. The structural proximity of tryptic fragments provided information about the structural arrangement of the hydrophilic protein domains, which was integrated into a three-dimensional model of CopA. Comparative modeling of CopA was guided by the sequence similarity to the calcium ATPase of the sarcoplasmic reticulum, Serca1, for which detailed structures are available. In addition, known partial structures of CPx-ATPase homologous to CopA were used as modeling templates. A docking approach was used to predict the orientation of the heavy metal binding domain of CopA relative to the core structure, which was verified by distance constraints derived from cross-links. The overall structural model of CopA resembles the Serca1 structure, but reveals distinctive features of CPx-type ATPases. A prominent feature is the positioning of the heavy metal binding domain. It features an orientation of the Cu binding ligands which is appropriate for the interaction with Cu-loaded metallochaperones in solution. Moreover, a novel model of the architecture of the intramembranous Cu binding sites could be derived.

A Food Recognition System for Diabetic Patients based on an Optimized Bag of Features Model

Computer vision-based food recognition could be used to estimate a meal's carbohydrate content for diabetic patients. This study proposes a methodology for automatic food recognition, based on the Bag of Features (BoF) model. An extensive technical investigation was conducted for the identification and optimization of the best performing components involved in the BoF architecture, as well as the estimation of the corresponding parameters. For the design and evaluation of the prototype system, a visual dataset with nearly 5,000 food images was created and organized into 11 classes. The optimized system computes dense local features, using the scale-invariant feature transform on the HSV color space, builds a visual dictionary of 10,000 visual words by using the hierarchical k-means clustering and finally classifies the food images with a linear support vector machine classifier. The system achieved classification accuracy of the order of 78%, thus proving the feasibility of the proposed approach in a very challenging image dataset.