A Novel Approach to Regularising 1NN classifier for Improved Generalization
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866911776280412160 |
|---|---|
| author | Challa, Aditya Danda, Sravan Najman, Laurent |
| author_facet | Challa, Aditya Danda, Sravan Najman, Laurent |
| contents | In this paper, we propose a class of non-parametric classifiers, that learn arbitrary boundaries and generalize well.
Our approach is based on a novel way to regularize 1NN classifiers using a greedy approach. We refer to this class of classifiers as Watershed Classifiers. 1NN classifiers are known to trivially over-fit but have very large VC dimension, hence do not generalize well. We show that watershed classifiers can find arbitrary boundaries on any dense enough dataset, and, at the same time, have very small VC dimension; hence a watershed classifier leads to good generalization.
Traditional approaches to regularize 1NN classifiers are to consider $K$ nearest neighbours. Neighbourhood component analysis (NCA) proposes a way to learn representations consistent with ($n-1$) nearest neighbour classifier, where $n$ denotes the size of the dataset. In this article, we propose a loss function which can learn representations consistent with watershed classifiers, and show that it outperforms the NCA baseline. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2402_08405 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | A Novel Approach to Regularising 1NN classifier for Improved Generalization Challa, Aditya Danda, Sravan Najman, Laurent Machine Learning In this paper, we propose a class of non-parametric classifiers, that learn arbitrary boundaries and generalize well. Our approach is based on a novel way to regularize 1NN classifiers using a greedy approach. We refer to this class of classifiers as Watershed Classifiers. 1NN classifiers are known to trivially over-fit but have very large VC dimension, hence do not generalize well. We show that watershed classifiers can find arbitrary boundaries on any dense enough dataset, and, at the same time, have very small VC dimension; hence a watershed classifier leads to good generalization. Traditional approaches to regularize 1NN classifiers are to consider $K$ nearest neighbours. Neighbourhood component analysis (NCA) proposes a way to learn representations consistent with ($n-1$) nearest neighbour classifier, where $n$ denotes the size of the dataset. In this article, we propose a loss function which can learn representations consistent with watershed classifiers, and show that it outperforms the NCA baseline. |
| title | A Novel Approach to Regularising 1NN classifier for Improved Generalization |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2402.08405 |