Merken

# Environmental thematic map and easy probabilistic estimation of a threshold exceed

#### Automatisierte Medienanalyse

## Diese automatischen Videoanalysen setzt das TIB|AV-Portal ein:

**Szenenerkennung**—

**Shot Boundary Detection**segmentiert das Video anhand von Bildmerkmalen. Ein daraus erzeugtes visuelles Inhaltsverzeichnis gibt einen schnellen Überblick über den Inhalt des Videos und bietet einen zielgenauen Zugriff.

**Texterkennung**–

**Intelligent Character Recognition**erfasst, indexiert und macht geschriebene Sprache (zum Beispiel Text auf Folien) durchsuchbar.

**Spracherkennung**–

**Speech to Text**notiert die gesprochene Sprache im Video in Form eines Transkripts, das durchsuchbar ist.

**Bilderkennung**–

**Visual Concept Detection**indexiert das Bewegtbild mit fachspezifischen und fächerübergreifenden visuellen Konzepten (zum Beispiel Landschaft, Fassadendetail, technische Zeichnung, Computeranimation oder Vorlesung).

**Verschlagwortung**–

**Named Entity Recognition**beschreibt die einzelnen Videosegmente mit semantisch verknüpften Sachbegriffen. Synonyme oder Unterbegriffe von eingegebenen Suchbegriffen können dadurch automatisch mitgesucht werden, was die Treffermenge erweitert.

Erkannte Entitäten

Sprachtranskript

00:01

because we start with the next session the because we start with an session I want to announce the and inform development meeting we want to do that later but this will take time so I would like to

00:28

announce the an informal development meeting after the last session so this should be at 6 30 PM

00:35

and because the informative means that people interested to accommodate stay here for a little while

00:43

and it's kind unusual when history of all we can collect the modules which have been developed and which are going to be presented here I think it should be then there might be a possibility to make a total concurrent contributions but I can also module and also the documentation of course this can be discussed later and another request to me what you think the chance for example to organize graph in the Chilean Italian language maybe or the father concrete and we've made up with people being interested later today so up to the last question which include it made a year for 1 thing so

01:33

welcome to the 2nd borns than which is dedicated to Mathematica Maple and techniques that use mainly to methodological I have recommended to all of speaker into the proper time which is around 20 minutes this presentation is by the legal and illegal gadget malleable and and the market the the from what including you know it presented by the movie could be added In the title environmental thematic map and you the probability to get the nation of its threefold exceeded ladies and gentlemen good morning to everybody my speech will be properly 2 main topics for which we have studied and implemented within the hour the 1st topic you that a new approaches can provide a new approach to the prediction of digital surface model from the observation that a 2nd topic is quite a related to the 1st because of the he said properties on you will solve the interface more there to provide the speaker classification OK what do they do that perfect model of digital more than generally peaking and they need the 2 coordinators of the field that that the they basically variable phenomenon like let's say up to them of the here were removed from and in particular with the 2003 model you that either by a about the automatic that and it is well known that that the prediction of what did not have the model uh belongs to the class of the field prediction problems that you a branch of to the peak and it is also well known that the pp Kelly and we can say that the effect of affecting the purity of a digital surface model actually the 1st 1 that you that the automation accurately and the 2nd 1 is the consistency of NATO that adopted the to work for the the the the without entering into details we can distinguish 2 main families of the predictions made by the 1st 1 this of course that that the new technique and within them we can remember being there will be translating the their conflict of determining need of hour that that they that impose began in pose the problem being solved on the field that without considering the observation of behavior and that the because they cannot perform a corrupt and modeling of the conditional always on the hotter and with this account

04:28

because the people because there are more complex to use them and that there are many technique about provided you properly you that the more accurate about the and that can be uh around with that with the spectral viral on input prevalent in their relations the classical example of orthographic need of these there when known you creating the difficult financial now we move in university in that they all information contained in 3 complement to complement around the the the complement of the view that the remaining peak complement and about the complement that the act complement within the observation is the observation that the user always tuples of the incorrectly but before that date we deem that field that we give the applicant that we can distinguish that need to complement that that is usually mathematical model at and that it will polynomial model that in In the light of the sample and the amount of the of the trend is the inherent in the red and line that the practical component can be graphically peak that took place from around the track and the and the the mathematically represented by a secondary order stationary processes that the other

05:49

prominent nice mathematical properties the property that the Davies for beginning that the young baby to uh meaning that a planet that official secondary order process can be completely described by the body of them that the the kind of of the correlation function we have 1 here and 1 of the big examples and that the that the deterministic component can be expected by apriori known polynomial model was good features our unknown the prediction that the are 2 simplifying the argument that the 3rd thing is that the empirical diagram modeling that the ideal that diagram of the observation to repeat that they heuristic diagram over there graphic prostate so that they could take that you they can be pretty sure that is performed according to a mean squared error between people and you the the you perform at that on the observation and then by using data if you make a diagram and by imposing debate known you know more than you have reading is and that meet other seems the quite theme park to use about there and that the main drawback that the that in order to correctly if you you make the fury become part of of the few that we needed to know a priori that that anything complement because that we can empirically estimated the value even believe that we can have a structure that that many people the from the innovation it is obviously impossible if we don't know the meaning of the component that would be the 1st thing we have tried that to propose a simplified approach it is very simplified the world there are 2 remarkable that laughter and that demonstrated to work quite well so we have here to don't about that we have a ripple taking effect that before the woman that reading them and that each data preprocessing basically both of these is that it the 1st thing that we can neglect the dates patient correlation between observations so and that these people to these we can approximated the observation covariance matrix as anything in there and that to this we can by the square meter of that if you make the air in the in the polynomial model the quiz features for union model that from the regional information and that the 2 most important we can evaluate the correctness of all of the thing of William and molded by the import feature that the only true if you mated graffiti that a I don't want to bore you with the computation that indicate that you that the implementation is made by a contracting than that in order to for him in modern Library each polynomial more than even if you make the NDE is evaluated by feature that the end we be the union model library they more

09:10

than that you show them that satisfy 5 the feature up we can that we can uh remove the kind of from the original observational so we have that approximately the trended observation we can use the the the trended observational to it made of empirical around and after that we can finally use that 40 general variation we can use that we you that reading tell to ready for the the the according to the

09:41

university and Europe although we had been implemented within the hour that we the effect of 3 common the 1st 1 is completely implemented by out there and that it indeed performs the operation of observation that during the 2nd and the theoretical model of the middle column find that you a cajun uh by which we can perform the empirical value around the nation

10:10

and their university and the yeah is uh obviously the algorithm as well on some data believe an example let it be they can take the view that the simulated we have been written that the that will behavior can be in there the that couldn't be a during the year theory theory the composed by detecting many people and the practical part of where the fact from the from taking care the that of the spots observations we had you it we had a hundred look to the observations from knowledge and we have you will that the uh the observation we that our algorithm to but that digital face more the comprised of between that they what he did not them predicted feed quite the that we have an error of less than an hour or a little later than 1 millimeter and the

11:09

and the deletion of 42 . 5 0 . 4 the 2nd optical might be achieved can use that all out and I use uh a digital the more there would be no that they a classification of particles what they mean for classification mean given that you get a very small then I want to identify the say that that the fate of the view that face model which belonged to when the thing in the class that you define it by any given by the area under the minimum and the maximum value the classic approaches in the classification is that before we even if it late from given a class the given that a generic set of that he has said that that belongs to the class if and only if each that they

11:59

you predicted value is that we been at the pretty much the clustering obviously the the

12:06

protein so simply because if we no the prediction accuracy because the completely neglect that make errors in the pretty so the the the idea is that the give an idea that freedom although we know that they tell you where you can derive from to so and that moreover we know that you about begin the provide not only the the that the small prediction but but also can be developed here if the model that the thereafter will for each uh gotta face model failed uh um contains the data prediction accuracy so we have added a simplified approach to take into account also

12:54

the the information when performing a classification the the movement below that we need that to people to be the 1st 1 is that the prediction error of the review that according to 1 norm and use we that there are many and the and the body and the but to date prediction value the 2nd alternative is that the prediction error out should in correlated so for each for each we had a probability distribution that you think of all the prediction value and that will length of the the report from the view of the about from the that I call that we can identify an equal to 0 that take into

13:40

account of the prediction error when that that's the kind of a family man on their work in both of 1st people to the there is a classic alter these uh arbitrary but reasonable the 2nd property of the poor who were around you use the wrong because it is well known that the pretty traps spatially correlated and sold on the prediction error but that you feel the need that working they alter because we don't know all the full covariance matrix of the prediction and also we if we called mold view that we cannot manage numerically under the the I want you to remember that I want that both of these are less simplified and then the classical people believe that these inequalities not only on the correlation

14:34

but also the error in the prediction OK so we can compute the for genetic the probability that that that that belongs to thing it got the simply by integrated by integrating that they use abortion function

14:52

we mean that they said that the entire and that we also compute the probability that they belong to 1 of being adaptation graph that that the the class that going from amino end meaning that you blasting the beginning of my

15:10

integrated with the the probability function we have stated that the algorithm uh we have children in the the RDF you about the authority of empowerment we have that that that the that the smaller than they you can model the stock it then you engage viewable for higher over here and we have the uh at the the annotation

15:37

class over a centimeter at 1st I show you what about provided by the deterministic approach obviously if choose that the deterministic approaches that range from higher models which satisfy condition predicted values of making uh the the baby because then they can meet and then they are but if we

16:08

chose it probably is the approach that we had a priori to the define the probability level at which we want to classify it said a that belonging to the nation class because we for each phase we can compute the probability to belong to the class that end in order to select the thing we had to decide what you there uh probability level for example we can't use it to classify depends on the application for any of the players will this will eventually have an a centimeter with probability at the end of the year to 4 or 5

16:46

per cent of the the the our only choice when we apply the probability level that heretofore think that thing that we obtain a true high enough that the 2 of them those provided by the deterministic approach and that obviously the care when the dead than the previous 1 when we choose a probability level 5 conclusion and 1 conclusion note the conclusion is that consideration Bayer probably that about that are equal to those provided by the deterministic approach when we choose a probability level that 50 per cent of and that we think that the the simplified the probabilistic approach partnership about the faculty and the women that probably the information on the media that have a small but I'm not completely at equal that is not about that simplified and then they come without provided by the classical approach moreover by adopting a probability level we can truly they're all manner people you want the much to we want the air in the classification and the the the the the term that they're true then that by e environmental conditions and environmental considerations on these considerations the proposed that people that are being implemented a a graph on Monday and will be available and for like

18:39

weapons years 2 little remark that could you

18:55

repeat the paper and think about this type of project and that they can command if I understood your right to do what opponent that polynomial good yes but the of God to compete every function that polynomial yes but if you know from you read that for example another function for example if I will mention that and I think it would be better to use the function not to use it you approach and could

19:38

I II and the thing is that you don't produce will that a european funded projects for the implementation of the GA it for you run meaning that taking into account the environmental considerations and in my a title presentation ideas from the life of the use about being that I can be on time for and what about the order the Common market and yet I think I know what you pay a relational what is that that that we complemented there will be in the field because they have no other information it's better to optimize the a priori information have about that our approach is that even when I don't know anything about the the the so I get I tried to extract from the observation and nowadays the and and in our common we have implemented only polynomial molded I can implement all also other models for for the of then the beach will from the Eureka our idea after that we have to implement something so we have implemented a polynomial models that implemented the unit and the mean of of the phase they put out because we can repeat the quantum of yes that could be seen that happen

21:12

and you implemented the bonded the body of Ramsar but gonna many phenomena in ecology and and and I feel that we downloaded the particularly loans and that the uh modeled by a number on the body of announcements and you would all levels and things like that and included critically include could support the European we implemented OK uh no I

21:45

want to be clear and that in our implementation in a year's only related to that and the that nation by feature based on that day in the world that uh the prediction is done by calling that you've got the program I don't want you to come up with and then we only implemented to got to choose important common goal achieved about the median graph and to do that by removing the by using so we we use the reduce library

22:23

if more a matter of questions OK but

00:00

Verbandstheorie

Vorlesung/Konferenz

Information

Softwareentwickler

00:22

Arithmetisches Mittel

Verbandstheorie

Information

Softwareentwickler

00:40

Soundverarbeitung

Graph

Kategorie <Mathematik>

Güte der Anpassung

Formale Sprache

Klasse <Mathematik>

Verzweigendes Programm

Familie <Mathematik>

Versionsverwaltung

Sprachsynthese

Kombinatorische Gruppentheorie

Modul

Mapping <Computergraphik>

Informationsmodellierung

Variable

Datenfeld

Prognoseverfahren

Luenberger-Beobachter

Vorlesung/Konferenz

Programmierumgebung

Widerspruchsfreiheit

Koordinaten

Schnittstelle

04:26

Prozess <Physik>

Kovarianzmatrix

Implementierung

Kartesische Koordinaten

Computerunterstütztes Verfahren

Computeranimation

Weg <Topologie>

Informationsmodellierung

Prognoseverfahren

Stichprobenumfang

Programmbibliothek

Luenberger-Beobachter

Zusammenhängender Graph

Datenstruktur

Gerade

Korrelationsfunktion

Parametersystem

Lineares Funktional

Sichtenkonzept

Kategorie <Mathematik>

Relativitätstheorie

Klassische Physik

Ideal <Mathematik>

Design by Contract

Arithmetisches Mittel

Polynom

Diagramm

Datenfeld

Twitter <Softwareplattform>

Information

Ordnung <Mathematik>

Fehlermeldung

Lesen <Datenverarbeitung>

09:09

Soundverarbeitung

Nichtlinearer Operator

TVD-Verfahren

Luenberger-Beobachter

Grundraum

Physikalische Theorie

Computeranimation

10:08

Arithmetisches Mittel

Informationsmodellierung

Sichtenkonzept

Algorithmus

Flächeninhalt

Extrempunkt

Klasse <Mathematik>

Mereologie

Meter

Luenberger-Beobachter

Partikelsystem

Grundraum

Physikalische Theorie

Fehlermeldung

11:58

Informationsmodellierung

Prognoseverfahren

Anwendungsspezifischer Prozessor

Fehlermeldung

12:54

Diskrete Wahrscheinlichkeitsverteilung

Dicke

Sichtenkonzept

Kategorie <Mathematik>

Kovarianzmatrix

Familie <Mathematik>

Computeranimation

Prognoseverfahren

Ungleichung

Äußere Algebra eines Moduls

Information

Normalvektor

Verkehrsinformation

Korrelationsfunktion

Metropolitan area network

Fehlermeldung

14:32

Arithmetisches Mittel

Lineares Funktional

Prognoseverfahren

Graph

Anpassung <Mathematik>

Klasse <Mathematik>

Computeranimation

Fehlermeldung

15:08

Autorisierung

Lineares Funktional

Informationsmodellierung

Algorithmus

Konditionszahl

Klasse <Mathematik>

Vorlesung/Konferenz

16:05

Graph

Klasse <Mathematik>

Fakultät <Mathematik>

Kartesische Koordinaten

Term

Computeranimation

Übergang

Konditionszahl

Hypermedia

Information

Ordnung <Mathematik>

Programmierumgebung

Phasenumwandlung

Auswahlaxiom

18:26

Lineares Funktional

Polynom

Rechter Winkel

Grundsätze ordnungsmäßiger Datenverarbeitung

Datentyp

Projektive Ebene

19:37

Videospiel

Besprechung/Interview

Zahlenbereich

Implementierung

Kombinatorische Gruppentheorie

Übergang

Informationsmodellierung

Einheit <Mathematik>

Datenfeld

Quantisierung <Physik>

Luenberger-Beobachter

Vorlesung/Konferenz

Projektive Ebene

Information

Ordnung <Mathematik>

Phasenumwandlung

21:43

Prognoseverfahren

Graph

Besprechung/Interview

Programmbibliothek

Implementierung

Vorlesung/Konferenz

Optimierung

Medianwert

Große Vereinheitlichung

### Metadaten

#### Formale Metadaten

Titel | Environmental thematic map and easy probabilistic estimation of a threshold exceed |

Serientitel | Open source GIS - GRASS user conference 2002 |

Anzahl der Teile | 45 |

Autor |
Biagi Ludovico, Negretti, Marco Maria, Antonia Politecnico, Milano di Brovelli, Maria |

Lizenz |
CC-Namensnennung - keine Bearbeitung 3.0 Deutschland: Sie dürfen das Werk in unveränderter Form zu jedem legalen Zweck nutzen, vervielfältigen, verbreiten und öffentlich zugänglich machen, sofern Sie den Namen des Autors/Rechteinhabers in der von ihm festgelegten Weise nennen. |

DOI | 10.5446/21764 |

Herausgeber | University of Trento |

Erscheinungsjahr | 2002 |

Sprache | Englisch |

#### Technische Metadaten

Dauer | 22:39 |

#### Inhaltliche Metadaten

Fachgebiet | Informatik |