For linear regression with up to 5000 samples and 50 features, the Moore-Penrose pseudoinverse outperforms batch gradient descent in speed and accuracy, especially on ill-conditioned data.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Comparing the Moore-Penrose Pseudoinverse and Gradient Descent for Solving Linear Regression Problems: A Performance Analysis
For linear regression with up to 5000 samples and 50 features, the Moore-Penrose pseudoinverse outperforms batch gradient descent in speed and accuracy, especially on ill-conditioned data.