Exercise_content an der Johannes Kepler Universität Linz | Karteikarten & Zusammenfassungen

Lernmaterialien für Exercise_content an der Johannes Kepler Universität Linz

Greife auf kostenlose Karteikarten, Zusammenfassungen, Übungsaufgaben und Altklausuren für deinen Exercise_content Kurs an der Johannes Kepler Universität Linz zu.

TESTE DEIN WISSEN

Best loss type for linear regression?

Lösung anzeigen
TESTE DEIN WISSEN

Quadratic loss

Lösung ausblenden
TESTE DEIN WISSEN

Is there a global solution in regression tasks?

Lösung anzeigen
TESTE DEIN WISSEN

Yes

Lösung ausblenden
TESTE DEIN WISSEN

Goal of using generalization error:

Lösung anzeigen
TESTE DEIN WISSEN

Minimizing the generalization error (learning) is concerned with optimizing bias and variance simultaneously.

Lösung ausblenden
TESTE DEIN WISSEN

Pitfall of Empirical risk minimization:

Lösung anzeigen
TESTE DEIN WISSEN

Empirical risk minimization does not include any mechanism to assess bias and variance independently (how should it?).

Lösung ausblenden
TESTE DEIN WISSEN

How to solve a dual problem?

Lösung anzeigen
TESTE DEIN WISSEN

1) Lagrange function

2) Karush-Kuhn-Tucker (KKT)

Lösung ausblenden
TESTE DEIN WISSEN

What is a good choise if we want linearly seperated data?

Lösung anzeigen
TESTE DEIN WISSEN

SVMs

Lösung ausblenden
TESTE DEIN WISSEN

What is a good choise if we want non-linearly seperated data?

Lösung anzeigen
TESTE DEIN WISSEN

C-SVMs

Lösung ausblenden
TESTE DEIN WISSEN

Linear separability is very restrictive. Where can it be achieved easier?

Lösung anzeigen
TESTE DEIN WISSEN

The higher the dimensionality, the easier linear separability can be achieved.

Lösung ausblenden
TESTE DEIN WISSEN

What learning algorithm and search type do all decision trees have?

Lösung anzeigen
TESTE DEIN WISSEN

All decision tree learning algorithms are recursive, depth-first search algorithms that perform hierarchical splits.

Lösung ausblenden
TESTE DEIN WISSEN

Which types of features are there when splitting in decision tree?

Lösung anzeigen
TESTE DEIN WISSEN

Categorical and numerical features

Lösung ausblenden
TESTE DEIN WISSEN

Standard splitting algorithm for classification:

Lösung anzeigen
TESTE DEIN WISSEN

Standard splitting criterion employed by the decision tree algorithm CART for classification.

Lösung ausblenden
TESTE DEIN WISSEN

How to estimate OOB-Error?

Lösung anzeigen
TESTE DEIN WISSEN

We take all trees that don't contain a certain datapoint. Then we take that datapoint and put it into all of those trees. Then we compute the predicted average over all those trees for this datapoint. We do that for all datapoints and take the average.

This is almost identical to cross validation.

Lösung ausblenden
  • 51218 Karteikarten
  • 946 Studierende
  • 50 Lernmaterialien

Beispielhafte Karteikarten für deinen Exercise_content Kurs an der Johannes Kepler Universität Linz - von Kommilitonen auf StudySmarter erstellt!

Q:

Best loss type for linear regression?

A:

Quadratic loss

Q:

Is there a global solution in regression tasks?

A:

Yes

Q:

Goal of using generalization error:

A:

Minimizing the generalization error (learning) is concerned with optimizing bias and variance simultaneously.

Q:

Pitfall of Empirical risk minimization:

A:

Empirical risk minimization does not include any mechanism to assess bias and variance independently (how should it?).

Q:

How to solve a dual problem?

A:

1) Lagrange function

2) Karush-Kuhn-Tucker (KKT)

Mehr Karteikarten anzeigen
Q:

What is a good choise if we want linearly seperated data?

A:

SVMs

Q:

What is a good choise if we want non-linearly seperated data?

A:

C-SVMs

Q:

Linear separability is very restrictive. Where can it be achieved easier?

A:

The higher the dimensionality, the easier linear separability can be achieved.

Q:

What learning algorithm and search type do all decision trees have?

A:

All decision tree learning algorithms are recursive, depth-first search algorithms that perform hierarchical splits.

Q:

Which types of features are there when splitting in decision tree?

A:

Categorical and numerical features

Q:

Standard splitting algorithm for classification:

A:

Standard splitting criterion employed by the decision tree algorithm CART for classification.

Q:

How to estimate OOB-Error?

A:

We take all trees that don't contain a certain datapoint. Then we take that datapoint and put it into all of those trees. Then we compute the predicted average over all those trees for this datapoint. We do that for all datapoints and take the average.

This is almost identical to cross validation.

Exercise_content

Erstelle und finde Lernmaterialien auf StudySmarter.

Greife kostenlos auf tausende geteilte Karteikarten, Zusammenfassungen, Altklausuren und mehr zu.

Jetzt loslegen

Das sind die beliebtesten Exercise_content Kurse im gesamten StudySmarter Universum

Content Management

Duale Hochschule Baden-Württemberg

Zum Kurs
Exercise Biology

TU München

Zum Kurs
Content Management

Universität Trier

Zum Kurs

Die all-in-one Lernapp für Studierende

Greife auf Millionen geteilter Lernmaterialien der StudySmarter Community zu
Kostenlos anmelden Exercise_content
Erstelle Karteikarten und Zusammenfassungen mit den StudySmarter Tools
Kostenlos loslegen Exercise_content