Robust LassoNet enhances neural network feature selection with robust loss functions

Researchers have introduced Robust LassoNet, an advancement on the existing LassoNet method for feature selection in neural networks. This new approach enhances the model's ability to handle noisy data by integrating robust loss functions like Huber, Cauchy, and Tukey's bisquare. Experiments demonstrate that Robust LassoNet improves both prediction accuracy and the precision of feature selection when dealing with outliers or heavy-tailed noise, while performing comparably on clean datasets. AI

IMPACT Improves the reliability of neural network feature selection in the presence of noisy data.

RANK_REASON The cluster contains a research paper detailing a new method for feature selection in neural networks. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv stat.ML →

• arXiv
• Cauchy loss
• Huber loss function
• MSE loss
• Robust LassoNet
• Tukey's bisquare loss

• paper
• model release

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →