Print Email Facebook Twitter Can Invariant Risk Minimization resist the temptation of learning spurious correlations? Title Can Invariant Risk Minimization resist the temptation of learning spurious correlations? Author van Lith, Jochem (TU Delft Electrical Engineering, Mathematics and Computer Science; TU Delft Pattern Recognition and Bioinformatics) Contributor Karlsson, R.K.A. (mentor) Bongers, S.R. (mentor) Krijthe, J.H. (mentor) Degree granting institution Delft University of Technology Programme Computer Science and Engineering Project CSE3000 Research Project Date 2022-06-24 Abstract Learning algorithms can perform poorly in unseen environments when they learnspurious correlations. This is known as the out-of-domain (OOD) generalization problem. Invariant Risk Minimization (IRM) is a method that attempts to solve this problem by learning invariant relationships. Motivating examples as well as counterexamples have been proposed about the performance of IRM. This work aims to clarify when the method works well and when it fails by testing its ability to learn invariant relationships. Therefore, experiments are done on a synthetic data model which simulates four data distribution shifts: covariate shift (CS), confounder based shift (CF), anti-causal shift (AC), and hybrid shift (HB). The experiments exploit IRM’s behaviour with respect to hetero- and homoskedasticity and adaptation of the training environments. We measure the error with regards to the optimal invariant predictor and compare to the non invariant Empirical Risk Minimization (ERM). The results show that IRM is generally able to learn invariance for the CS and CF shifts, especially when the deviation between the training environments is large. In the AC and HB shifts, this strongly depends on the values of the training environments. Subject invariance principleMachine learningGeneralization To reference this document use: http://resolver.tudelft.nl/uuid:60339154-302c-407a-9047-05d9b3b21f57 Part of collection Student theses Document type bachelor thesis Rights © 2022 Jochem van Lith Files PDF Final_Paper.pdf 1.87 MB Close viewer /islandora/object/uuid:60339154-302c-407a-9047-05d9b3b21f57/datastream/OBJ/view