GLORIA

GEOMAR Library Ocean Research Information Access

X
feed icon rss

Your email was sent successfully. Check your inbox.

An error occurred while sending the email. Please try again.

Proceed reservation?

Export
  • 1
    facet.materialart.
    Unknown
    Supervised learning (2021)
    Elsevier B.V.
    Details
    Publication Date: 2021-02-01
    Description: In supervised classification, we search criteria allowing us to decide whether a sample belongs to a certain class of patterns. The identification of such decision functions is based on examples where we know a priori to which class they belong. The distinction of seismic signals, produced from earthquakes and nuclear explosions, is a classical problem of discrimination using classification with supervision. We move on from observed data—signals originating from known earthquakes and nuclear tests—and search for criteria on how to assign a class to a signal of unknown origin. We begin with Principal Component Analysis (PCA) and Fisher's Linear Discriminant Analysis (FLDA), identifying a linear element separating groups at best. PCA, FLDA, and likelihood-based approaches make use of statistical properties of the groups. Considering only the number of misclassified samples as a cost, we may prefer alternatives, such as the Multilayer Perceptrons (MLPs). The Support Vector Machines (SVMs) use a modified cost function, combining the criterion of the minimum number of misclassified samples with a request of separating the hulls of the groups with a margin as wide as possible. Both SVMs and MLPs overcome the limits of linear discrimination. A famous example for the advantages of the two techniques is the eXclusive OR (XOR) problem, where we wish to form classes of objects having the same parity—even, e.g., (0,0), (1,1) or odd, e.g., (0,1), (1,0). MLPs and SVMs offer effective methods for the identification of nonlinear decision functions, allowing us to resolve classification problems of any complexity provided the data set used during earning is sufficiently large. In Hidden Markov Models (HMMs), we consider observations where their meaning depends on their context. Observations form a causal chain generated by a hidden process. In Bayesian Networks (BNs) we represent conditional (in)dependencies between a set of random variables by a graphical model. In both HMMs and BNs, we aim at identifying models and parameters that explain observations with a highest possible degree of probability.
    Description: Published
    Description: 33-85
    Description: 5T. Sismologia, geofisica e geologia per l'ingegneria sismica
    Keywords: pattern recognition ; supervised learning ; Support Vector Machines ; Multilayer Perceptrons ; Hidden Markov Models ; Bayesian Networks ; 04.04. Geology ; 04.06. Seismology ; 04.07. Tectonophysics ; 04.08. Volcanology ; 05.04. Instrumentation and techniques of general interest
    Repository Name: Istituto Nazionale di Geofisica e Vulcanologia (INGV)
    Type: book chapter
    Location Call Number Limitation Availability
    BibTip Others were also interested in ...
    PAPER (OA)
  • 2
    facet.materialart.
    Unknown
    Patterns, objects, and features (2021)
    Elsevier B.V.
    Details
    Publication Date: 2021-02-01
    Description: Patterns and objects are described by a variety of characteristics, namely features and feature vectors. Features can be numerical, ordinal, and categorical. Patterns can be made up of a number of objects, such as in speech processing. In geophysics, numerical features are the most common ones and we focus on those. The choice of appropriate features requires a priori reasoning about the physical relation between patterns and features. We present strategies for feature identification and procedures suitable for pattern recognition. In time series analysis and image processing, the direct use of raw data is not feasible. Procedures of feature extraction, based on locally encountered characteristics of the data, are applied. Here we present the problem of delineating segments of interest in time series and textures in image processing. In transformations, we “translate” our raw data to a form suitable for learning. In Principal Component Analysis, we rotate the original features to a system of uncorrelated variables, limiting redundancy. Independent Component Analysis follows a similar strategy, transforming our data into variables independent of each other. Fourier transform and wavelet transform are based on the representation of the original data as a series of basis functions—sines and cosines or finite-length wavelets. Redundancy reduction is achieved considering the contributions of the single basis functions. Even though a large number of features help to solve a classification problem, feature vectors with high dimensions pose severe problems. Besides the computational burden, we encounter problems known under the term “curse of dimensionality.” The curse of dimensionality entails the necessity of feature selection and reduction, which includes a priori considerations as well as redundancy reduction. The significance of features may be evaluated with tests, such as Student’s t or Hotelling's T2, and, in more complex problems, with cross-validation methods.
    Description: Published
    Description: 3-13
    Description: 5T. Sismologia, geofisica e geologia per l'ingegneria sismica
    Keywords: pattern recognition ; objects ; features ; 04.04. Geology ; 04.06. Seismology ; 04.07. Tectonophysics ; 04.08. Volcanology ; 05.04. Instrumentation and techniques of general interest
    Repository Name: Istituto Nazionale di Geofisica e Vulcanologia (INGV)
    Type: book chapter
    Location Call Number Limitation Availability
    BibTip Others were also interested in ...
    PAPER (OA)
  • 3
    facet.materialart.
    Unknown
    Integrated monitoring of soil gases, plume SO2 and volcanic tremor to detect impulsive magma transfer at Mt. Etna volcano (Italy) (2020)
    EGU
    Details
    Publication Date: 2020-10-13
    Description: Magma transfer in an open-conduit volcano is a complex process that is still open to debate and not entirely understood. For this reason, a multidisciplinary monitoring of active volcanoes is not only welcome, but also necessary for a correct comprehension of how volcanoes work. Mt. Etna is probably one of the best test sites for doing this, because of the large multidisciplinary monitoring network setup by the Osservatorio Etneo of Istituto Nazionale di Geofisica e Vulcanologia (INGV-OE), the high frequency of eruptions and the relatively easy access to most of its surface. We present new data on integrated monitoring of volcanic tremor, plume sulphur dioxide (SO2) flux and soil hydrogen (H2) and carbon dioxide (CO2) concentration from Mt. Etna. The RMS amplitude of volcanic tremor was measured by seismic stations at various distances from the summit craters, plume SO2 flux was measured from nine stations around the volcano and soil gases were measured in a station located in a low-temperature (T ∼ 85 °C) fumarole field on the upper north side of the volcano. During our monitoring period, we observed clear and marked anomalous changes in all parameters, with a nice temporal sequence that started with a soil CO2 and SO2 flux increase, followed a few days later by a soil H2 spike-like increase and finally with sharp spike-like increases in RMS amplitude (about 24 h after the onset of the anomaly in H2) at all seismic stations. After the initial spikes, all parameters returned more or less slowly to their background levels. Geochemical data, however, showed persistence of slight anomalous degassing for some more weeks, even in the apparent absence of RMS amplitude triggers. This suggests that the conditions of slight instability in the degassing magma column inside the volcano conduits lasted for a long period, probably until return to some sort of balance with the “normal” pressure conditions. The RMS amplitude increase accompanied the onset of strong Strombolian activity at the Northeast Crater, one of the four summit craters of Mt. Etna, which continued during the following period of moderate geochemical anomalies. This suggests a cause-effect relationship between the anomalies observed in all parameters and magma migration inside the central conduits of the volcano. Volcanic tremor is a well-established key parameter in the assessment of the probability of eruptive activity at Etna and it is actually used as a basis for a multistation system for detection of volcanic anomalies that has been developed by INGV-OE at Etna. Adding the information provided by our geochemical parameters gave us more solid support to this system, helping us understand better the mechanisms of magma migration inside of an active, open-conduit basaltic volcano.
    Description: Published
    Description: online (due to Covid pandemic)
    Description: 4V. Processi pre-eruttivi
    Keywords: integrated monitoring ; soil gases ; plume SO2 ; volcanic tremor ; magma transfer ; Etna ; 04.08. Volcanology
    Repository Name: Istituto Nazionale di Geofisica e Vulcanologia (INGV)
    Type: Oral presentation
    Location Call Number Limitation Availability
    BibTip Others were also interested in ...
    PAPER (OA)
  • 4
    facet.materialart.
    Unknown
    Massive data collection in volcanic areas owing to photogrammetryderived models: a key example from the NE Rift, Mt Etna (Italy). (2021)
    Details
    Publication Date: 2021-07-16
    Description: The collection of a conspicuous amount of data in volcanic areas is a key for a deeper understanding of the relationships between faulting, diking and superficial volcanic processes. A way to quickly collect huge amounts of data is to analyse photogrammetry-derived models (Digital surface models, orthomosaics and 3D models) using Unmanned Aerial Vehicles (UAVs) to collect all necessary pictures obtaining final models with a texture ground resolution up to 2-3 cm/pix. In this work, we describe our approach to build up models of a broad area located in the NE Rift of Mt. Etna, which is affected by continuous ground deformation linked to gravity sliding of the eastern flank of the volcano and dyke injection. The area is characterized by the presence of eruptive craters and fissures, extension fractures, and normal faults, as well as by historical lava flows. The goal was to quantify the kinematics at extensional fractures and normal faults, integrating the latter with seismological data to reconstruct the stress field acting in this peculiar sector of the volcano. By the point of view of UAV surveying, the test area is challenging since it is located at an altitude ranging between 2700 and 1900 m a.s.l., and it is affected by extreme weather conditions, like a strong wind. Resulting models, in the form of DSM and orthomosaic, are characterised by a resolution of 11.86 and 2.97 cm/pix, respectively, obtained from the elaboration of 4018 photos and covering an area of 2.2 km2. Thanks to these models, we recognized the presence of 20 normal fault segments, 250 extension fractures, and 54 single eruptive fissures. Considering all the above mention data, we quantified the kinematics at extensional fractures and normal faults, obtaining an extension rate of 1.9 cm/yr for the last 406 yr.
    Description: INGV
    Description: Published
    Description: Vienna
    Description: 2V. Struttura e sistema di alimentazione dei vulcani
    Keywords: Etna ; Drone ; SfM tecniques ; NE rift ; 04.08. Volcanology
    Repository Name: Istituto Nazionale di Geofisica e Vulcanologia (INGV)
    Type: Conference paper
    Location Call Number Limitation Availability
    BibTip Others were also interested in ...
    PAPER (OA)
  • 5
    facet.materialart.
    Unknown
    A-posteriori Analyses of Pattern Recognition Results (2020)
    EGU
    Details
    Publication Date: 2020-10-13
    Description: Data-driven approaches applied to to large and complex data sets are intriguing, however the results must be revised with a critical attitude. For example, a diagnostic tool may provide hints for a serious disease, or for anomalous conditions potentially indicating an impending natural risk. The demand of a high score of identified anomalies – true positives - comes together with the request of a low percentage of false positives. Indeed, a high rate of false positives can ruin the diagnostics. Receiver Operation Curves (ROC) allows us to find a reasonable compromise between the need of accuracy of the diagnostics and robustness with respect to false alerts. In multiclass problems success is commonly measured as the score for which calculated and target classification of patterns matches at best. A high score does not automatically mean that a method is truly effective. Its value becomes questionable, when a random guess leads to a high score as well. The so called “Kappa Statistics” is an elegant way to assess the quality of a classification scheme. We present some case studies demonstrating how such a-posteriori analysis helps corroborate the results. Sometimes an approach does not lead to the desired success. In thes cases, a sound a-posteriori analysis of the reasons for the failure often provide interesting insights into the problem, Those problems may reside in an inappropriate definition of the targets, inadequate features, etc. Often the problems can be fixed just by adjusting some choices. Finally, a change of strategy may be necessary in order to achieve a more satisfying result. In the applications presented here, we highlight the pitfalls arising in particular from ill-defined targets and unsuitable feature selections. The validation of unsupervised learning is still a matter of debate. Some formal criteria (e. g. Davies Bouldin Index, Silhouette Index or other) are available for centroid-based clustering where a unique metric valid for all clusters can be defined. Difficulties arise when metrics are defined individually for each single cluster (for instance, Gaussian Model clusters, adaptive criteria) as well as using schemes where centroids are essentially meaningless. This is the case in density based clustering. In all these cases, users are better off when asking themselves whether a clustering is meaningful for the problem in physical terms. In our presentation we discuss the problem of choosing a suitable number of clusters in cases in which formal criteria are not applicable. We demonstrate how the identification of groups of patterns helps the identification of elements which have a clear physical meaning, even when strict rules for assessing the clustering are not available.
    Description: Published
    Description: online (for the Covid pandemic)
    Description: 3IT. Calcolo scientifico
    Keywords: pattern recognition ; machine learning ; statistics ; data processing ; 05.06. Methods ; 05.01. Computational geophysics
    Repository Name: Istituto Nazionale di Geofisica e Vulcanologia (INGV)
    Type: Poster session
    Location Call Number Limitation Availability
    BibTip Others were also interested in ...
    PAPER (OA)
  • 6
    facet.materialart.
    Unknown
    Augmented reality for volcanic and seismic risk communication (2020)
    EGU
    Details
    Publication Date: 2020-10-13
    Description: Dealing with topics concerning natural risk management in a volcanic environment, can greatly benefit from innovative techniques. In particular, Augmented Reality (AR) and Virtual Reality (VR) are well known by Native Digital and can be used by lower-level and university students to promote their understanding of natural risks. 3DTeLC is a three-year trans-European project funded by the Erasmus+ Key Action 2 programme: “Cooperation for Innovation and Exchange of Good Practices, a European scheme that fosters higher education partnerships” (https://www.erasmusplus.org.uk/key-action-2). The main goal of this project is to help young students to become highly-skilled professionals in the field of environment and geosciences, gaining knowledge in image and 3D-spatial analysis, data management and informatics, and strengthening their mathematical and numerical skills in Earth observation and data analysis. In the framework of this project INGV team has developed a “Talking poster”, using a custom AR tool to propose a user friendly approach aimed at the reduction of volcanic and seismic risks.
    Description: Published
    Description: online (for the Covid pandemic)
    Description: 1TM. Formazione
    Keywords: augmented reality ; natural risk management ; education ; innovative techniques ; 05.03. Educational, History of Science, Public Issues ; 05.06. Methods ; 05.08. Risk
    Repository Name: Istituto Nazionale di Geofisica e Vulcanologia (INGV)
    Type: Oral presentation
    Location Call Number Limitation Availability
    BibTip Others were also interested in ...
    PAPER (OA)
  • 7
    facet.materialart.
    Unknown
    A posteriori analyses—advantages and pitfalls of pattern recognition techniques (2021)
    Elsevier B.V.
    Details
    Publication Date: 2021-02-01
    Description: In this chapter, we deal with a posterior analysis of supervised and unsupervised learning techniques. Concerning supervised learning, we discuss methods of cross-validation and assessment of uncertainty of tests by means of the “Receiver Operation Curve” and the “Kappa-Statistics.” We show the importance of appropriate target information. Furthermore, features are critical; when they are not properly chosen, they fail to describe objects in a unique way. A critical attitude is mandatory to validate the success of an application. A high score of success does not automatically mean that a method is truly effective. At the same time, users should not despair when the desired success is not achieved. A posteriori analysis on the reasons for an apparent failure may provide useful insights into the problem. Targets may not be appropriately defined, features can be inadequate, etc. Problems can be often fixed by adjusting a few choices; sometimes a change of strategy may be necessary to improve results. In unsupervised learning, we ask whether the structures revealed in the data are meaningful. Cluster analysis offers rules giving formal answers to this question; however, such rules are not generally applicable. In some cases, a heuristic approach may be necessary.
    Description: Published
    Description: 237-259
    Description: 5T. Sismologia, geofisica e geologia per l'ingegneria sismica
    Keywords: pattern recognition ; a posteriori analysis ; supervised learning ; unsupervised learning ; cross validation ; assessment of uncertainty ; Receiver Operation Curve ; Kappa-Statistics ; 04.04. Geology ; 04.06. Seismology ; 04.07. Tectonophysics ; 04.08. Volcanology ; 05.04. Instrumentation and techniques of general interest
    Repository Name: Istituto Nazionale di Geofisica e Vulcanologia (INGV)
    Type: book chapter
    Location Call Number Limitation Availability
    BibTip Others were also interested in ...
    PAPER (OA)
  • 8
    facet.materialart.
    Unknown
    Unsupervised learning (2021)
    Elsevier B.V.
    Details
    Publication Date: 2021-02-01
    Description: Unsupervised learning is based on the definition of an appropriate metrics defining the similarity of patterns. On the basis of the metrics, we form groups or clusters of patterns following various strategies. In partitioning cluster analysis, we form disjoint clusters. Being faced with data, where clusters still exhibit heterogeneities or subclusters, we may adopt the strategy of hierarchical clustering, which leads to the generation of the so-called dendrograms. In the partitioning strategy, we choose a priori the number of clusters we wish to form, whereas in the hierarchical strategy, the number of clusters depends on the resolution we want to have. Density-based clustering considers local structures of a data set. We consider a unit volume in our data space and derive the density of samples within this volume. Moving toward neighboring volumes, we verify whether the number of samples has dropped below a threshold. If this is the case, we identify a heterogeneity, otherwise we join the neighboring volumes to a common cluster. Self-Organizing Maps (SOMs) provide a way of representing multidimensional data in much lower dimensional spaces than the original data set. The process of reducing the dimensionality of vectors is essentially a data compression technique known as vector quantization. The SOM technique creates a network that stores information in a way that it maintains the topological relationships within the patterns of the data set. Each node of the network represents a number of patterns. Assigning a color code to the nodes, the representation of pattern characteristics with high-dimensional feature vectors becomes extremely effective.
    Description: Published
    Description: 87-124
    Description: 5T. Sismologia, geofisica e geologia per l'ingegneria sismica
    Keywords: pattern recognition ; unsupervised learning ; cluster analysis ; Density-based clustering ; Self-Organizing Maps ; 04.04. Geology ; 04.06. Seismology ; 04.07. Tectonophysics ; 04.08. Volcanology ; 05.04. Instrumentation and techniques of general interest
    Repository Name: Istituto Nazionale di Geofisica e Vulcanologia (INGV)
    Type: book chapter
    Location Call Number Limitation Availability
    BibTip Others were also interested in ...
    PAPER (OA)
  • 9
    facet.materialart.
    Unknown
    Applications with unsupervised learning (2021)
    Elsevier B.V.
    Details
    Publication Date: 2021-02-01
    Description: This chapter demonstrates how Unsupervised Learning can be applied in Geophysics. It starts with an example of clustering seismic spectra obtained on Stromboli volcano. K-means clustering as well as clustering using the Adaptive Criterion are applied. The latter criterion is preferred as it better matches the statistical characteristics of the data. Clusters show close relation to the state of volcanic activity. Density based clustering reveals groups whose hulls can be of irregular shape. This makes the method attractive, among others, for the identification of structural elements in geology, which often do not have a simple geometry. An example application is discussed considering the distribution of earthquake locations on Mt Etna, which clearly evidence structures already identified by other, independent evidences. Using SOM we aim at data reduction and effective graphical visualization. In an example for climate data we demonstrate the application of SOM for zoning purposes. Besides, the temporal evolution of spectral seismic data recorded on Mt Etna can be effectively monitored using SOM. We further illustrate the use of SOM for directional data, which can be handled best using a toroidal sheet geometry. We discuss this using a data set of seismic moment tensors of Mediterranean earthquakes.
    Description: Published
    Description: 189-234
    Description: 5T. Sismologia, geofisica e geologia per l'ingegneria sismica
    Keywords: pattern recognition ; unsupervised learning ; Density based clustering ; Stromboli ; earthquakes ; volcanic activity ; structural data ; seismic moment tensors ; 04.04. Geology ; 04.06. Seismology ; 04.07. Tectonophysics ; 04.08. Volcanology ; 05.04. Instrumentation and techniques of general interest
    Repository Name: Istituto Nazionale di Geofisica e Vulcanologia (INGV)
    Type: book chapter
    Location Call Number Limitation Availability
    BibTip Others were also interested in ...
    PAPER (OA)
  • 10
    facet.materialart.
    Unknown
    Software manuals (2021)
    Elsevier B.V.
    Details
    Publication Date: 2021-02-01
    Description: In this chapter, we present scripts and programs that accompany this book. Five MATLAB scripts regard simple examples related to supervised learning, that is, linear discrimination, the perceptron, support vector machines, and hidden Markov models. Seven scripts are devoted to unsupervised learning, such as K-means and fuzzy clustering, agglomerative clustering, density-based clustering, and clustering of patterns where features are correlated. These scripts provide a starting point for the reader, who can adjust and modify the codes with respect to proper needs. Besides, we provide sources and executables of programs that can be readily applied to larger and more complex datasets. These programs regard supervised learning using multilayerperceptron and support vector machines. KKAnalysis is a toolbox for unsupervised learning and offers various options of clustering and the use of self-organizing maps. The programs offer graphical user interfaces (GUI) to facilitate their use and create both graphical and alphanumeric output that can be used in further processing steps. The programs come along with real-world datasets that are also discussed in the example applications presented in various chapters of the book. Other propaedeutic material can be found in a folder called “miscellaneous.”
    Description: Published
    Description: 261-313
    Description: 5T. Sismologia, geofisica e geologia per l'ingegneria sismica
    Keywords: pattern recognition ; software manuals ; MATLAB scripts ; 04.04. Geology ; 04.06. Seismology ; 04.07. Tectonophysics ; 04.08. Volcanology ; 05.04. Instrumentation and techniques of general interest
    Repository Name: Istituto Nazionale di Geofisica e Vulcanologia (INGV)
    Type: book chapter
    Location Call Number Limitation Availability
    BibTip Others were also interested in ...
    PAPER (OA)
Close ⊗
This website uses cookies and the analysis tool Matomo. More information can be found here...