Precipitates segmentation from scanning electron microscope images through machine learning techniques

The presence of precipitates in metallic materials affects its durability, resistance and mechanical properties. Hence, its automatic identification by image processing and machine learning techniques may lead to reliable and efficient assessments on the materials. In this paper, we introduce four widely used supervised pattern recognition techniques to accomplish metallic precipitates segmentation in scanning electron microscope images from dissimilar welding on a Hastelloy C-276 alloy: Support Vector Machines, Optimum-Path Forest, Self Organizing Maps and a Bayesian classifier. Experimental results demonstrated that all classifiers achieved similar recognition rates with good results validated by an expert in metallographic image analysis. © 2011 Springer-Verlag Berlin Heidelberg.

Particle competition and cooperation to prevent error propagation from mislabeled data in semi-supervised learning

Semi-supervised learning is applied to classification problems where only a small portion of the data items is labeled. In these cases, the reliability of the labels is a crucial factor, because mislabeled items may propagate wrong labels to a large portion or even the entire data set. This paper aims to address this problem by presenting a graph-based (network-based) semi-supervised learning method, specifically designed to handle data sets with mislabeled samples. The method uses teams of walking particles, with competitive and cooperative behavior, for label propagation in the network constructed from the input data set. The proposed model is nature-inspired and it incorporates some features to make it robust to a considerable amount of mislabeled data items. Computer simulations show the performance of the method in the presence of different percentage of mislabeled data, in networks of different sizes and average node degree. Importantly, these simulations reveals the existence of the critical points of the mislabeled subset size, below which the network is free of wrong label contamination, but above which the mislabeled samples start to propagate their labels to the rest of the network. Moreover, numerical comparisons have been made among the proposed method and other representative graph-based semi-supervised learning methods using both artificial and real-world data sets. Interestingly, the proposed method has increasing better performance than the others as the percentage of mislabeled samples is getting larger. © 2012 IEEE.

Learning HMMs for nucleotide sequences from amino acid alignments

Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)

Active Learning to Maximize Area Under the ROC Curve

In active learning, a machine learning algorithmis given an unlabeled set of examples U, and is allowed to request labels for a relatively small subset of U to use for training. The goal is then to judiciously choose which examples in U to have labeled in order to optimize some performance criterion, e.g. classification accuracy. We study how active learning affects AUC. We examine two existing algorithms from the literature and present our own active learning algorithms designed to maximize the AUC of the hypothesis. One of our algorithms was consistently the top performer, and Closest Sampling from the literature often came in second behind it. When good posterior probability estimates were available, our heuristics were by far the best.

Modal hedges in para-pharmaceutical product instructions: some examples from English and Serbian

[EN] The paper investigates how modal hedges (Coates 1983) understood as expressions of procedural meaning , i.e. expressions which instruct the addressee/reader how to process the propositional content of an utterance/statement (Watts 2004) are used in product descriptions, advertisements and consumer instructions leaflets for a number of products belonging to the Consumer Health Care category for the purposes of complying with consumer protection laws on the one hand and serving as an implicit disclaimer of manufacturer’s responsibility on the other. The analysis is carried out contrastively for two languages, English and Serbian. The results obtained are discussed and viewed as a matter of cultural variety and difference, especially taking into consideration the fact that consumer protection laws seem to be equally strict in US, UK and Commonwealth, Europe and Serbia.

Tectonics and kinematics of curved mountain belts: examples from the Andes and the Alps

Curved mountain belts have always fascinated geologists and geophysicists because of their peculiar structural setting and geodynamic mechanisms of formation. The need of studying orogenic bends arises from the numerous questions to which geologists and geophysicists have tried to answer to during the last two decades, such as: what are the mechanisms governing orogenic bends formation? Why do they form? Do they develop in particular geological conditions? And if so, what are the most favorable conditions? What are their relationships with the deformational history of the belt? Why is the shape of arcuate orogens in many parts of the Earth so different? What are the factors controlling the shape of orogenic bends? Paleomagnetism demonstrated to be one of the most effective techniques in order to document the deformation of a curved belt through the determination of vertical axis rotations. In fact, the pattern of rotations within a curved belt can reveal the occurrence of a bending, and its timing. Nevertheless, paleomagnetic data alone are not sufficient to constrain the tectonic evolution of a curved belt. Usually, structural analysis integrates paleomagnetic data, in defining the kinematics of a belt through kinematic indicators on brittle fault planes (i.e., slickensides, mineral fibers growth, SC-structures). My research program has been focused on the study of curved mountain belts through paleomagnetism, in order to define their kinematics, timing, and mechanisms of formation. Structural analysis, performed only in some regions, supported and integrated paleomagnetic data. In particular, three arcuate orogenic systems have been investigated: the Western Alpine Arc (NW Italy), the Bolivian Orocline (Central Andes, NW Argentina), and the Patagonian Orocline (Tierra del Fuego, southern Argentina). The bending of the Western Alpine Arc has been investigated so far using different approaches, though few based on reliable paleomagnetic data. Results from our paleomagnetic study carried out in the Tertiary Piedmont Basin, located on top of Alpine nappes, indicate that the Western Alpine Arc is a primary bend that has been subsequently tightened by further ~50° during Aquitanian-Serravallian times (23-12 Ma). This mid-Miocene oroclinal bending, superimposing onto a pre-existing Eocene nonrotational arc, is the result of a composite geodynamic mechanism, where slab rollback, mantle flows, and rotating thrust emplacement are intimately linked. Relying on our paleomagnetic and structural evidence, the Bolivian Orocline can be considered as a progressive bend, whose formation has been driven by the along-strike gradient of crustal shortening. The documented clockwise rotations up to 45° are compatible with a secondary-bending type mechanism occurring after Eocene-Oligocene times (30-40 Ma), and their nature is probably related to the widespread shearing taking place between zones of differential shortening. Since ~15 Ma ago, the activity of N-S left-lateral strike-slip faults in the Eastern Cordillera at the border with the Altiplano-Puna plateau induced up to ~40° counterclockwise rotations along the fault zone, locally annulling the regional clockwise rotation. We proposed that mid-Miocene strike-slip activity developed in response of a compressive stress (related to body forces) at the plateau margins, caused by the progressive lateral (southward) growth of the Altiplano-Puna plateau, laterally spreading from the overthickened crustal region of the salient apex. The growth of plateaux by lateral spreading seems to be a mechanism common to other major plateaux in the Earth (i.e., Tibetan plateau). Results from the Patagonian Orocline represent the first reliable constraint to the timing of bending in the southern tip of South America. They indicate that the Patagonian Orocline did not undergo any significant rotation since early Eocene times (~50 Ma), implying that it may be considered either a primary bend, or an orocline formed during the late Cretaceous-early Eocene deformation phase. This result has important implications on the opening of the Drake Passage at ~32 Ma, since it is definitely not related to the formation of the Patagonian orocline, but the sole consequence of the Scotia plate spreading. Finally, relying on the results and implications from the study of the Western Alpine Arc, the Bolivian Orocline, and the Patagonian Orocline, general conclusions on curved mountain belt formation have been inferred.

The problems of equivalence in advertising translation: examples from English and Russian

Abstract The academic environment has recently recognized the importance and benefits that an extensive research on the translation of advertising can have for translation studies. Despite the growing interest and increasing research activity in the field it is still difficult to speak about a theory of advertising translation in general. There is a need for further study encompassing different languages and both heterogeneous and homogenous cultures that will give the possibility to receive a more complete map of what the translation of advertising is and should be. Previous studies have been concentrated, for the most part, on Western European language pairs. This study is a research into perfume and cosmetics print advertisements translated from English into Russian where both visual and verbal elements are considered. Three broad translation approaches have been identified in what concerns the verbal message: Translated message, parallel translation, recreated adverts, and three approaches in dealing with the image: similar images, modified images, completely different images. The thesis shows that where Russian advertisements for perfume products tend to have a message, or create one, this is often lacking in the English copy. The article ends by suggesting that perfume advertisements favor the standardization approach when entering Russian market. The attempts to localize the advert have also been noticed although they are obviously less numerous in perfume adverts and are rather instances of adaptation - a mix between the localization and standardization approaches since they keep drawing on the same globally accepted universals about female beauty and concern for ‘woman’s identity’ (we focused our analysis on products designed for female consumers). This study, complementing previous studies, aims to be a contribution to the description of laws and strategies that guide the translation of advertising texts into Russian.

Integrating new and traditional approaches for the estimate of slip-rates of active faults: examples from the Mw 6.3, 2009 L’Aquila earthquake area, Central Italy

This thesis is based on the integration of traditional and innovative approaches aimed at improving the normal faults seimogenic identification and characterization, focusing mainly on slip-rate estimate as a measure of the fault activity. The L’Aquila Mw 6.3 April 6, 2009 earthquake causative fault, namely the Paganica - San Demetrio fault system (PSDFS), was used as a test site. We developed a multidisciplinary and scale‐based strategy consisting of paleoseismological investigations, detailed geomorphological and geological field studies, as well as shallow geophysical imaging and an innovative application of physical properties measurements. We produced a detailed geomorphological and geological map of the PSDFS, defining its tectonic style, arrangement, kinematics, extent, geometry and internal complexities. The PSDFS is a 19 km-long tectonic structure, characterized by a complex structural setting and arranged in two main sectors: the Paganica sector to the NW, characterized by a narrow deformation zone, and the San Demetrio sector to SE, where the strain is accommodated by several tectonic structures, exhuming and dissecting a wide Quaternary basin, suggesting the occurrence of strain migration through time. The integration of all the fault displacement data and age constraints (radiocarbon dating, optically stimulated luminescence (OSL) and tephrochronology) helped in calculating an average Quaternary slip-rate representative for the PSDFS of 0.27 - 0.48 mm/yr. On the basis of its length (ca. 20 km) and slip per event (up to 0.8 m) we also estimated a max expected Magnitude of 6.3-6.8 for this fault. All these topics have a significant implication in terms of surface faulting hazard in the area and may contribute also to the understanding of the PSDFS seismic behavior and of the local seismic hazard.

Lattice preferred orientation in omphacite: Examples from the Tauern Window, Austria and the Western Gneiss Region, Norway

It is lively debated how eclogites find their way from deep to mid-crustal levels during exhumation. Different exhumation models for high-pressure and ultrahigh-pressure rocks were suggested in previous studies, based mainly on field observations and less on microstructural studies on the exhumed rocks. The development and improvement of electron microscopy techniques allows it, to focus interest on direct investigations of microstructures and crystallographic properties in eclogites. In this case, it is of importance to study the applicability of crystallographic measurements on eclogites for exhumation processes and to unravel which processes affect eclogite textures. Previous studies suggested a strong relationship between deformation and lattice preferred orientation (LPO) in omphacite but it is still unclear if the deformation is related to the exhumation of eclogites. This study is focused on the questions which processes affect omphacite LPO and if textural investigations of omphacite are applicable for studying eclogite exhumation. Therefore, eclogites from two examples in the Alps and in the Caledonides were collected systematically and investigated with respect to omphacite LPO by using the electron backscattered diffraction (EBSD) technique. Omphacite textures of the Tauern Window (Austria) and the Western Gneiss Region (Norway) were studied to compare lattice preferred orientation with field observations and suggested exhumation models from previous studies. The interpretation of omphacite textures, regarding the deformation regime is mainly based on numerical simulations in previous studies. Omphacite LPO patterns of the Eclogite Zone are clearly independent from any kind of exhumation process. The textures were generated during omphacite growth on the prograde path of eclogite development until metamorphic peak conditions. Field observations in the Eclogite Zone show that kinematics in garnet mica schist, surrounding the eclogites, strongly indicate an extrusion wedge geometry. Stretching lineations show top-N thrusting at the base and a top-S normal faulting with a sinistral shear component at the top of the Eclogite Zone. The different shear sense on both sides of the unit does not affect the omphacite textures in any way. The omphacite lattice preferred orientation patterns of the Western Gneiss Region can not be connected with any exhumation model. The textures were probably generated during the metamorphic peak and reflect the change from subduction to exhumation. Eclogite Zone and Western Gneiss Region differ significantly in size and especially in metamorphic conditions. While the Eclogite Zone is characterized by constant P-T conditions (600-650°C, 20-25 kbar), the Western Gneiss Region contains a wide P-T range from high- to ultrahigh pressure conditions (400-800°C, 20-35 kbar). In contrast to this, the omphacite textures of both units are very similar. This means that omphacite LPO is independent from P-T conditions and therefore from burial depth. Further, in both units, omphacite LPO is independent from grain and subgrain size as well as from any shape preferred orientation (SPO) on grain and subgrain scale. Overall, omphacite lattice preferred orientation are generated on the prograde part of omphacite development. Therefore, textural investigations on omphacite LPO are not applicable to study eclogite exhumation.

Transgenic mice as a tool for depression research: examples from the endocannabinoid and the corticotropin-releasing hormone system

Major depression belongs to the most serious and widespread psychiatric disorders in today’s society. There is a great need for the delineation of the underlying molecular mechanisms as well as for the identification of novel targets for its treatment. In this thesis, transgenic mice of the endocannabinoid and the corticotropin-releasing hormone (CRH) system were investigated to determine the putative role of these systems for depression-like phenotypes in mice. In the first part of the thesis, we found that the endocannabinoid system was prominently involved in a brain region-specific and temporally controlled manner in acute as well as in chronic stress processing. Genetic deletion in combination with pharmacological intervention revealed the importance of a fully functional endocannabinoid system for efficient neuroendocrine and behavioral stress coping. Accordingly, cannabinoid type 1 (CB1) receptor-deficient mice displayed several depression-like symptoms and molecular alterations, including “behavioral despair”, stress hormone hypersecretion and decreased glucocorticoid receptor and brain-derived neurotrophic factor expression in the hippocampus. However, the endocannabinoid system was dispensable for the efficacy of currently used antidepressant drugs. To facilitate future endocannabinoid research, a transgenic mouse was generated, which overexpressed the CB1 receptor protein fused to a fluorescent protein. In the second part of the thesis, conditional brain region-specific CRH overexpressing mice were evaluated as a model for pathological chronic CRH hyperactivation. Mutant mice showed aberrant neuroendocrine and behavioral stress coping and hyperarousal due to CRH-induced activation of the noradrenergic system in the brain. Mutant mice appeared to share similarities with naturally occurring endogenous CRH activation in wild-type mice and were sensitive to acute pharmacological blockade of CRH receptor type 1 (CRH-R1). Thus, CRH overexpressing mice serve as an ideal in vivo tool to evaluate the efficacy of novel CRH-R1 antagonists. Together, these findings highlight the potential of transgenic mice for the understanding of certain endo-phenotypes (isolated symptoms) of depression and their molecular correlates.

Spatio-temporal credit assignment in population learning

Learning by reinforcement is important in shaping animal behavior, and in particular in behavioral decision making. Such decision making is likely to involve the integration of many synaptic events in space and time. However, using a single reinforcement signal to modulate synaptic plasticity, as suggested in classical reinforcement learning algorithms, a twofold problem arises. Different synapses will have contributed differently to the behavioral decision, and even for one and the same synapse, releases at different times may have had different effects. Here we present a plasticity rule which solves this spatio-temporal credit assignment problem in a population of spiking neurons. The learning rule is spike-time dependent and maximizes the expected reward by following its stochastic gradient. Synaptic plasticity is modulated not only by the reward, but also by a population feedback signal. While this additional signal solves the spatial component of the problem, the temporal one is solved by means of synaptic eligibility traces. In contrast to temporal difference (TD) based approaches to reinforcement learning, our rule is explicit with regard to the assumed biophysical mechanisms. Neurotransmitter concentrations determine plasticity and learning occurs fully online. Further, it works even if the task to be learned is non-Markovian, i.e. when reinforcement is not determined by the current state of the system but may also depend on past events. The performance of the model is assessed by studying three non-Markovian tasks. In the first task, the reward is delayed beyond the last action with non-related stimuli and actions appearing in between. The second task involves an action sequence which is itself extended in time and reward is only delivered at the last action, as it is the case in any type of board-game. The third task is the inspection game that has been studied in neuroeconomics, where an inspector tries to prevent a worker from shirking. Applying our algorithm to this game yields a learning behavior which is consistent with behavioral data from humans and monkeys, revealing themselves properties of a mixed Nash equilibrium. The examples show that our neuronal implementation of reward based learning copes with delayed and stochastic reward delivery, and also with the learning of mixed strategies in two-opponent games.

Ramachandran-type Plots for Glycosidic Linkages: Examples from Molecular Dynamics Simulations using the Glycam06 Force Field

The goals of this article are to (1) provide further validation of the Glycam06 force field, specifically for its use in implicit solvent molecular dynamic (MD) simulations, and (2) to present the extension of G.N. Ramachandran's idea of plotting amino acid phi and psi angles to the glycosidic phi, psi, and omega angles formed between carbohydrates. As in traditional Ramachandran plots, these carbohydrate Ramachandran-type (carb-Rama) plots reveal the coupling between the glycosidic angles by displaying the allowed and disallowed conformational space. Considering two-bond glycosidic linkages, there are 18 possible conformational regions that can be defined by (α, ϕ, ψ) and (β, ϕ, ψ), whereas for three-bond linkages, there are 54 possible regions that can be defined by (α, ϕ, ψ, ω) and (β, ϕ, ψ, ω). Illustrating these ideas are molecular dynamic simulations on an implicitly hydrated oligosaccharide (700 ns) and its eight constituent disaccharides (50 ns/disaccharide). For each linkage, we compare and contrast the oligosaccharide and respective disaccharide carb-Rama plots, validate the simulations and the Glycam06 force field through comparison to experimental data, and discuss the general trends observed in the plots.