On the other hand, in the presence of the spurious feature, the full model can fit the training data perfectly with a smaller norm by assigning weight \(1\) for the feature \(s\) (\(|<\theta^\text<-s>>|_2^2 = 4\) while \(|<\theta^\text<+s>>|_2^2 + w^2 = 2 < 4\)).
Generally, in the overparameterized regime, since the number of training examples is less than the number of features, there are some directions of data variation that are not observed in the training data. In this example, we do not observe any information about the second and third features. However, the non-zero weight for the spurious feature leads to a different assumption for the unseen directions. In particular, the full model does not assign weight \(0\) to the unseen directions. Indeed, by substituting \(s\) with \(<\beta^\star>^\top z\), we can view the full model as not using \(s\) but implicitly assigning weight \(\beta^\star_2=2\) to the second feature and \(\beta^\star_3=-2\) to the third feature (unseen directions at training).
Inside analogy, removing \(s\) reduces the error eros escort Dallas TX to possess a test shipping with high deviations from no towards the next function, while removing \(s\) increases the error to possess an examination delivery with a high deviations out of zero for the third element.
As we saw in the previous example, by using the spurious feature, the full model incorporates \(<\beta^\star>\) into its estimate. The true target parameter (\(\theta^\star\)) and the true spurious feature parameters (\(<\beta^\star>\)) agree on some of the unseen directions and do not agree on the others. Thus, depending on which unseen directions are weighted heavily in the test time, removing \(s\) can increase or decrease the error.
More formally, the weight assigned to the spurious feature is proportional to the projection of \(\theta^\star\) on \(<\beta^\star>\) on the seen directions. If this number is close to the projection of \(\theta^\star\) on \(<\beta^\star>\) on the unseen directions (in comparison to 0), removing \(s\) increases the error, and it decreases the error otherwise. Note that since we are assuming noiseless linear regression and choose models that fit training data, the model predicts perfectly in the seen directions and only variations in unseen directions contribute to the error.
(Left) The new projection out of \(\theta^\star\) into \(\beta^\star\) is actually positive on seen guidance, however it is bad regarding the unseen advice; therefore, removing \(s\) reduces the mistake. (Right) The fresh new projection out-of \(\theta^\star\) towards the \(\beta^\star\) is similar both in viewed and you will unseen instructions; therefore, removing \(s\) boosts the mistake.
Let’s now formalize the conditions under which removing the spurious feature (\(s\)) increases the error. Let \(\Pi = Z(ZZ^\top)^<-1>Z\) denote the column space of training data (seen directions), thus \(I-\Pi\) denotes the null space of training data (unseen direction). The below equation determines when removing the spurious feature decreases the error.
The latest leftover top is the difference in brand new projection off \(\theta^\star\) towards the \(\beta^\star\) regarding seen direction with regards to projection in the unseen advice scaled by test date covariance. Ideal side is the difference in 0 (we.e., not using spurious has) while the projection of \(\theta^\star\) for the \(\beta^\star\) from the unseen recommendations scaled because of the shot date covariance. Deleting \(s\) support if for example the remaining top are more than best front side.
Since concept can be applied just to linear models, we currently reveal that inside low-linear activities coached on the genuine-community datasets, deleting a spurious function reduces the precision and you may has an effect on communities disproportionately.