Karteikarten
Linear and Dual Regression
Karten in diesem Stapel (6)
- 1Linear regression modelling guidelines?1) Define model y = w^To(x) + e
2) Our assumption of addative noise and that it has a specific distribution leads to likely
3) Specify prior and get posterior from Gaussian identities - 2In linear regression what is the prediction?We want to see what the updated posterior says about new data
We take the weighted average of all possible weights w and see what the value is at x* and weigh this average by how likely we think the specific w is. - 3Why would we use dual linear regression?Maps a non-linear problem to a different space where it's linear
- 4How to do dual linear regression?Formulate posterior
Find stationary point of posterior
Rewrite w in terms of the data
Kernel regression
Now we can relate new data to old data - 5Kernel regression equation and explanation
- 6Problems with dual linear regression?Lost uncertainty to gain adaptability
Informationen
- Beschreibung
- Machine Learning Karteikarten am Linear and Dual Regression, erstellt von Philip Masters am 03/01/2020.
- Erstellt von
- Philip Masters · Jan 2020
- Aktualisiert
- 3 Jan 2020
- Aufrufe
- 2
- Bewertung