Noise misleads rotation invariant algorithms on sparse targets
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2024
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866910354040160256 |
|---|---|
| author | Warmuth, Manfred K. Kotłowski, Wojciech Jones, Matt Amid, Ehsan |
| author_facet | Warmuth, Manfred K. Kotłowski, Wojciech Jones, Matt Amid, Ehsan |
| contents | It is well known that the class of rotation invariant algorithms are suboptimal even for learning sparse linear problems when the number of examples is below the "dimension" of the problem. This class includes any gradient descent trained neural net with a fully-connected input layer (initialized with a rotationally symmetric distribution). The simplest sparse problem is learning a single feature out of $d$ features. In that case the classification error or regression loss grows with $1-k/n$ where $k$ is the number of examples seen. These lower bounds become vacuous when the number of examples $k$ reaches the dimension $d$.
We show that when noise is added to this sparse linear problem, rotation invariant algorithms are still suboptimal after seeing $d$ or more examples. We prove this via a lower bound for the Bayes optimal algorithm on a rotationally symmetrized problem. We then prove much lower upper bounds on the same problem for simple non-rotation invariant algorithms. Finally we analyze the gradient flow trajectories of many standard optimization algorithms in some simple cases and show how they veer toward or away from the sparse targets.
We believe that our trajectory categorization will be useful in designing algorithms that can exploit sparse targets and our method for proving lower bounds will be crucial for analyzing other families of algorithms that admit different classes of invariances. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2403_02697 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Noise misleads rotation invariant algorithms on sparse targets Warmuth, Manfred K. Kotłowski, Wojciech Jones, Matt Amid, Ehsan Machine Learning It is well known that the class of rotation invariant algorithms are suboptimal even for learning sparse linear problems when the number of examples is below the "dimension" of the problem. This class includes any gradient descent trained neural net with a fully-connected input layer (initialized with a rotationally symmetric distribution). The simplest sparse problem is learning a single feature out of $d$ features. In that case the classification error or regression loss grows with $1-k/n$ where $k$ is the number of examples seen. These lower bounds become vacuous when the number of examples $k$ reaches the dimension $d$. We show that when noise is added to this sparse linear problem, rotation invariant algorithms are still suboptimal after seeing $d$ or more examples. We prove this via a lower bound for the Bayes optimal algorithm on a rotationally symmetrized problem. We then prove much lower upper bounds on the same problem for simple non-rotation invariant algorithms. Finally we analyze the gradient flow trajectories of many standard optimization algorithms in some simple cases and show how they veer toward or away from the sparse targets. We believe that our trajectory categorization will be useful in designing algorithms that can exploit sparse targets and our method for proving lower bounds will be crucial for analyzing other families of algorithms that admit different classes of invariances. |
| title | Noise misleads rotation invariant algorithms on sparse targets |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2403.02697 |