Implicit Bias in Matrix Factorization and its Explicit Realization in a New Architecture

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Hou, Yikun, Sra, Suvrit, Yurtsever, Alp
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866912684254953472
author Hou, Yikun
Sra, Suvrit
Yurtsever, Alp
author_facet Hou, Yikun
Sra, Suvrit
Yurtsever, Alp
contents Gradient descent for matrix factorization exhibits an implicit bias toward approximately low-rank solutions. While existing theories often assume the boundedness of iterates, empirically the bias persists even with unbounded sequences. This reflects a dynamic where factors develop low-rank structure while their magnitudes increase, tending to align with certain directions. To capture this behavior in a stable way, we introduce a new factorization model: $X\approx UDV^\top$, where $U$ and $V$ are constrained within norm balls, while $D$ is a diagonal factor allowing the model to span the entire search space. Experiments show that this model consistently exhibits a strong implicit bias, yielding truly (rather than approximately) low-rank solutions. Extending the idea to neural networks, we introduce a new model featuring constrained layers and diagonal components that achieves competitive performance on various regression and classification tasks while producing lightweight, low-rank representations.
format Preprint
id arxiv_https___arxiv_org_abs_2501_16322
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Implicit Bias in Matrix Factorization and its Explicit Realization in a New Architecture
Hou, Yikun
Sra, Suvrit
Yurtsever, Alp
Machine Learning
Optimization and Control
Gradient descent for matrix factorization exhibits an implicit bias toward approximately low-rank solutions. While existing theories often assume the boundedness of iterates, empirically the bias persists even with unbounded sequences. This reflects a dynamic where factors develop low-rank structure while their magnitudes increase, tending to align with certain directions. To capture this behavior in a stable way, we introduce a new factorization model: $X\approx UDV^\top$, where $U$ and $V$ are constrained within norm balls, while $D$ is a diagonal factor allowing the model to span the entire search space. Experiments show that this model consistently exhibits a strong implicit bias, yielding truly (rather than approximately) low-rank solutions. Extending the idea to neural networks, we introduce a new model featuring constrained layers and diagonal components that achieves competitive performance on various regression and classification tasks while producing lightweight, low-rank representations.
title Implicit Bias in Matrix Factorization and its Explicit Realization in a New Architecture
topic Machine Learning
Optimization and Control
url https://arxiv.org/abs/2501.16322