Summary of A Semi-supervised Learning Using Over-parameterized Regression, by Katsuyuki Hagiwara
A semi-supervised learning using over-parameterized regression
by Katsuyuki Hagiwara
First submitted to arxiv on: 6 Sep 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Methodology (stat.ME)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The paper proposes a novel approach to semi-supervised learning (SSL) in regression problems, where a few labeled samples and many unlabeled samples are available. The method incorporates information from the unlabeled samples into kernel functions using Gaussian kernels with centers as input samples. This leads to an over-parameterized regression problem, which is tackled by applying minimum norm least squares (MNLS). The MNLS is used for feature extraction/dimension reduction in the SVD representation of a Gram type matrix. Several methods are proposed based on this framework, including thresholding according to singular value magnitude with cross-validation and universal thresholding. Experimental results show that these SVD regression methods outperform traditional ridge regression methods on real datasets. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Semi-supervised learning is a way for computers to learn from some labeled data and lots of unlabeled data. In this paper, scientists come up with a new approach to use the unlabeled data in a special kind of problem called regression. They use something called Gaussian kernels that connect labeled and unlabeled data together. This makes the problem too hard to solve normally, so they use a different method called minimum norm least squares. They also propose some new ways to use this method for feature extraction and dimension reduction. The results show that their approach works better than the traditional way on real datasets. |
Keywords
» Artificial intelligence » Feature extraction » Regression » Semi supervised
