1,721,140 research outputs found
Metric Learning for Prototype-Based Classification
Full text linkBiehl M, Hammer B, Schneider P, Villmann T. Metric learning for prototype based classification. In: Bianchini M, Maggini M, Scarselli F, eds. Innovations in Neural Information – Paradigms and Applications. Studies in Computational Intelligence, 247. Berlin: Springer; 2009: 183-199
Recursive Neural Networks and Automata
No full textRecursive neural networks are computational models that can be used to pro- cess structured data. In particular recurrent neural networks can be trained to process temporal sequences following the rules embedded in a set of examples. Their computation relies on a set of real-valued state and input variables which are processed using a set of continuous operators such as multipliers, adders, and sigmoidal functions. Despite of their strictly continuous nature, many researchers have also investigated interesting symbolic aspects related to their continuous nonlinear dynamics (see e.g. [1-3]), trying to understand the relationships be- tween these models and the classical symbolic computational models, like Turing machines and Finite-State Automata (FSA). The use of real-valued variables seems to provide these models with an unbounded quantity of memory capacity whose effective use is mainly restricted by the functions used to update them and to compute the output sequence
The schizosphere-plastosphere boundary through rheological modelling across and along a fold-and-thrust belt: case studies from the Hellenides, Greece, and seismotectonic applications
No full textIn this work we used rheological modelling, by means of strength profiles calibration and realization, to determine the depth of the brittle-ductile transition in the Aegean Region, more specifically in relationship with the Hellenides fold and thrust belt. The final aim is to compare the rheological results with the corresponding seismicity and apply the results to seismotectonic issues. Indeed the brittle-ductile transition, though not exactly coinciding with, represents a fair approximation to the seismic-aseismic boundary, which is more precisely related to the velocity weakening-velocity strengthening transition (Tse and Rice 1986; Scholz 1988). During the interseismic period we can, in any case, consider the depth of the brittleductile transition (hereinafter BDT) as a reasonable indicator for the maximum depth extent of (seismogenic) faulting. Accordingly, a precise rheological modelling can help distinguishing the brittle, potentially seismogenic layer, also termed schizosphere (after Scholz 1988), from the undelying, ductilely deforming plastosphere. We then decided to apply the rheological modelling to the Aegean Region for a twofold reason: firstly, it represents one of the most seismically active area all over the world, and therefore it allows to compare and validate the rheological results with a totally independent data source, namely well-located seismicity. Secondly, the Aegean Region is characterized by a dense pattern of seismogenic faults, whose terminations at depths are not always well defined and therefore represent ideal candidates for applying the rheological modelling results, in order to help constraining the geometry and the seismogenic potential of such faults, with the final aim of improving their relative seismotectonic characterization
Acoustic guidance of unmanned underwater vehicles for the execution of cooperative robotic tasks
No full textRheological behaviour in continental and oceanic subduction: Inferences for the seismotectonics of the Aegean region
No full textWe reconstructed several rheological transects across the Aegean Region, comparing the behaviour in collisional versus subducting settings. We interpolated closely spaced 1D strength envelopes, realized through a dedicated MATLAB script, for determining the shallow lithospheric distribution of brittle and ductile layers. We mainly used literature data and geodynamic considerations to fix the parameters for the rheological modelling and took particular care in reproducing reliable thermal models. The results of the mechanical-rheological model highlighted the following features and differences between the northern continental collision and the southern oceanic subduction settings: i) a slightly shallower brittle-ductile transition (BDT) in the western sectors of the northern transects (~30–33 km) with respect to the southern ones (~40 km); ii) on the contrary, in the central-eastern sectors of the investigated area, corresponding to an extensional tectonic regime, the northern transects have a relatively deeper BDT (about 20–25 km) compared with the southern ones (about 15 km); iii) the occurrence of a thick, deeper brittle layer below the shallowest BDT, in the central-eastern sectors of the northern transects. We suggest that such regional differences are mainly related and attributable to the surface heat flow distribution (which directly affects the geothermal gradient) and to the tectonic and geodynamic context. The results of the rheological modelling in terms of depth extent of the brittle layer(s) have been compared with the depth distribution of available relocated seismicity, showing good agreement with the rheological layering proposed here. Finally, the depth of the shallowest BDT along the transects has been adopted as a constraint for the seismogenic layer thickness. Such information has been used to improve the seismotectonic characterization of selected crustal seismogenic sources crossing the transects, by estimating their maximum potential magnitudes on the basis of their geometrical features and consistency with the rheological layering
Seismological data versus rheological modelling: Comparisons across the Aegean Region for improving the seismic hazard assessment
No full textCompared analyses of the seismicity cutoff depth and the corresponding BDT (brittle-ductile transition) depth obtained from rheological modelling have been performed for numerous test sites all over the Aegean Region. The major goal of this research is to determine whether the rheological transition could be effectively correlated to the seismological one, so that both information can independently contribute in determining the seismogenic layer thickness and hence in constraining the maximum width of active faults affecting a region. As concerns the seismological data, only relocated events have been considered for the purpose of this paper, included either in seismic sequences associated with a mainshock, or in datasets of background seismicity. In the same areas, we carried out rheological modelling for reconstructing the local strength envelope, focusing our attention on the BDT depth. The systematic comparison at the investigated sites indicates that 90% of the relocated seismicity always occurs within the corresponding rheological transition. Moreover, for the datasets also providing magnitudes, 99% of the total seismic scalar moment is always released above the BDT depth. Such results are verified for different tectonic regimes and geological settings, both extensional, transcurrent and compressional. The achieved results confirm that the BDT depth based on accurate strength envelopes could be considered a reliable indicator for the thickness of the seismogenic layer, thus contributing to constrain the maximum widths for active faults affecting those regions. Based on these constraints, we also applied some empirical relationships to estimate the possible maximum magnitude for seismogenic sources in correspondence of the tested sites, thus potentially improving our seismic hazard assessment analyses for these areas
Sensitivity analysis for crustal rheological profiles: Examples from the Aegean region
No full textWe tested the validity of rheological profiles and their sensitivity to variations in the input parameters, with particular emphasis on the resulting BDT (brittle-ductile transition) depth and corresponding strength and temperature. For this purpose, we selected two test-sites from the Aegean Region, one in an extensional tectonic setting and the other in a strike-slip regime, and carefully realized the corresponding “reference” rheological profiles based on literature data and specific geological constraints. The reference envelopes have been then compared with a set of different profiles realized by varying the input parameters of the constitutive equations of the brittle and ductile behaviours within reasonable ranges. Firstly, tests were performed by changing the value of only one input parameter per time, with the aim of quantifying and comparing its influence on the BDT properties. The parameters exerting the greatest control are the activation energy, the power-law exponent and the surface heat flow (through its influence on the geothermal gradient), for the creep behaviour. As regards the brittle behaviour parameters, the friction coefficient and the pore fluid pressure could play a significant role especially in determining the maximum strength. In a second phase, we simultaneously varied all the input parameters in order to consider the possible synergistic effects on the resulting rheological profiles and to verify the likelihood and consistency of the reference models. For the statistical approach, one hundred thousand random combinations of the analysed parameters have been generated. The particular care spent on selecting the range of values of each parameter is reflected in the results of the statistical analyses, which show a good agreement with the reference profiles and allow estimating the overall uncertainties. Finally, the obtained strength envelopes have been compared with the accurately relocated depth distribution of recent seismic sequences that affected the two test areas. In both cases, the depth corresponding to the 95% of the total released energy nicely fits the BDT depth obtained from the rheological modelling, therefore confirming that this parameter could represent a reasonable and reliable approximation of the seismogenic layer thickness
A Multi-sequence Alignment Algorithm for Web Template Detection
No full textNowadays most of Web pages are automatically assembled by content management systems or editing tools that apply a fixed template to give a uniform structure to all the documents beloging to the same site. The tem- plate usually contains side information that provides better graphics, navigation bars and menus, banners and advertisements that are aimed to improve the users’ browsing experience but may hinder tools for automatic processing of Web documents. In this paper, we present a novel template removing technique that exploits a sequence alignment algorithm from bioinformatics that is able to automatically extract the template from a quite small sample of pages from the same site. The algorithm detects the common structure of HTML tags among pairs of pages and merges the partial hypotheses using a binary tree consensus schema. The experi- mental results show that the algorithm is able to attain a good precision and recall in the retrieval of the real template structure exploiting just 16 sample pages from the site. Moreover, the positive impact of the template removing technique is shown on a Web page clustering task
- …
