Machine Learning at TU München | Flashcards & Summaries

Lernmaterialien für Machine Learning an der TU München

Greife auf kostenlose Karteikarten, Zusammenfassungen, Übungsaufgaben und Altklausuren für deinen Machine Learning Kurs an der TU München zu.

TESTE DEIN WISSEN
What is supervised learning?
Lösung anzeigen
TESTE DEIN WISSEN
Training samples, with their corresponding targets
--> find a function f that generalizes this relationship!
--> using f, make test predictions a different set of test data
Lösung ausblenden
TESTE DEIN WISSEN
What is Classification?
Lösung anzeigen
TESTE DEIN WISSEN
A subclass of supervised learning
--> the targets represent certain categories
Lösung ausblenden
TESTE DEIN WISSEN
What is Regression?
Lösung anzeigen
TESTE DEIN WISSEN
A subclass of supervised learning
--> The targets represent continuous numbers!
Lösung ausblenden
TESTE DEIN WISSEN
What are typical tasks in unsupervised learning?
Lösung anzeigen
TESTE DEIN WISSEN
  • Clustering
  • Dimensionality Reduction
  • Generative Modeling
  • Topic Models
Much more but likely not necessary for the exam
Lösung ausblenden
TESTE DEIN WISSEN
What is unsupervised learning?
Lösung anzeigen
TESTE DEIN WISSEN
Find structure in unlabeled data
Typical tasks are: 
  • Clustering
  • Dimensionality reduction
Lösung ausblenden
TESTE DEIN WISSEN
What is one of the first things to do when you try to analyse new data?
Lösung anzeigen
TESTE DEIN WISSEN
Visualize them (scattee plots, histograms, ...)
Lösung ausblenden
TESTE DEIN WISSEN
What is the hard part of decision trees?
Lösung anzeigen
TESTE DEIN WISSEN
Learning them. Using them is easy and much faster than KNNs for example.

Building the optimal tree is NP-complete!
Lösung ausblenden
TESTE DEIN WISSEN
How can we decide how many neighbours are best for our KNN?
Lösung anzeigen
TESTE DEIN WISSEN
Split your data into training, validation and Test Set!
Goal: generalize and pick k so that KNN performs best for unseen data
1. Learn model using the training set
2. Evaluate performance for different ks using the Validation set
3. Report final performance on the test set
Lösung ausblenden
TESTE DEIN WISSEN
How can we improve the generalization of a 1NN?
Lösung anzeigen
TESTE DEIN WISSEN
Look at the k-nearest-neighbours & majority vote --> KNN
Lösung ausblenden
TESTE DEIN WISSEN
What is the F1 Score?
Lösung anzeigen
TESTE DEIN WISSEN
A score to measure the performance of our model.
The harmonic mean of recall and precision

f1 = (2 * prec * rec)/(prec + rec)
Lösung ausblenden
TESTE DEIN WISSEN
What is the intuitive core idea of decision trees?
Lösung anzeigen
TESTE DEIN WISSEN
20-Question-Game
Lösung ausblenden
TESTE DEIN WISSEN
What can we do instead of just testing all possible combinationd of a decision tree?
Lösung anzeigen
TESTE DEIN WISSEN
Use greedy, local decisions. What is the best single local decision I can do at this point?
Lösung ausblenden
  • 356460 Karteikarten
  • 8257 Studierende
  • 335 Lernmaterialien

Beispielhafte Karteikarten für deinen Machine Learning Kurs an der TU München - von Kommilitonen auf StudySmarter erstellt!

Q:
What is supervised learning?
A:
Training samples, with their corresponding targets
--> find a function f that generalizes this relationship!
--> using f, make test predictions a different set of test data
Q:
What is Classification?
A:
A subclass of supervised learning
--> the targets represent certain categories
Q:
What is Regression?
A:
A subclass of supervised learning
--> The targets represent continuous numbers!
Q:
What are typical tasks in unsupervised learning?
A:
  • Clustering
  • Dimensionality Reduction
  • Generative Modeling
  • Topic Models
Much more but likely not necessary for the exam
Q:
What is unsupervised learning?
A:
Find structure in unlabeled data
Typical tasks are: 
  • Clustering
  • Dimensionality reduction
Mehr Karteikarten anzeigen
Q:
What is one of the first things to do when you try to analyse new data?
A:
Visualize them (scattee plots, histograms, ...)
Q:
What is the hard part of decision trees?
A:
Learning them. Using them is easy and much faster than KNNs for example.

Building the optimal tree is NP-complete!
Q:
How can we decide how many neighbours are best for our KNN?
A:
Split your data into training, validation and Test Set!
Goal: generalize and pick k so that KNN performs best for unseen data
1. Learn model using the training set
2. Evaluate performance for different ks using the Validation set
3. Report final performance on the test set
Q:
How can we improve the generalization of a 1NN?
A:
Look at the k-nearest-neighbours & majority vote --> KNN
Q:
What is the F1 Score?
A:
A score to measure the performance of our model.
The harmonic mean of recall and precision

f1 = (2 * prec * rec)/(prec + rec)
Q:
What is the intuitive core idea of decision trees?
A:
20-Question-Game
Q:
What can we do instead of just testing all possible combinationd of a decision tree?
A:
Use greedy, local decisions. What is the best single local decision I can do at this point?
Machine Learning

Erstelle und finde Lernmaterialien auf StudySmarter.

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

Jetzt loslegen

Das sind die beliebtesten StudySmarter Kurse für deinen Studiengang Machine Learning an der TU München

Für deinen Studiengang Machine Learning an der TU München gibt es bereits viele Kurse, die von deinen Kommilitonen auf StudySmarter erstellt wurden. Karteikarten, Zusammenfassungen, Altklausuren, Übungsaufgaben und mehr warten auf dich!

Das sind die beliebtesten Machine Learning Kurse im gesamten StudySmarter Universum

Machine Learning & Energy

TU Darmstadt

Zum Kurs
Machine Learning and Forecasting

Maastricht University

Zum Kurs
machine learning

Université des Sciences et de la Technologie Houari Boumediène

Zum Kurs

Die all-in-one Lernapp für Studierende

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