Summary of Semi-supervised Learning For Robust Speech Evaluation, by Huayun Zhang et al.
Semi-supervised Learning For Robust Speech Evaluation
by Huayun Zhang, Jeremy H.M. Wong, Geyu Lin, Nancy F. Chen
First submitted to arxiv on: 23 Sep 2024
Categories
- Main: Artificial Intelligence (cs.AI)
- Secondary: None
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 This paper proposes a novel approach to automatic speech evaluation that addresses common challenges in training models for this task. The authors recognize that traditional corpora used for training these models often suffer from sparsity and imbalanced distribution of scores, making them unreliable when faced with under-represented or out-of-distribution samples. To tackle this issue, they develop a semi-supervised pre-training method that leverages normalized mutual information to quantify speech characteristics and an interpolated loss function that minimizes both prediction error and divergence between two probability distributions. Experimental results on a public dataset show that this approach not only achieves high performance but also produces evenly distributed predictions across distinct proficiency levels, outperforming state-of-the-art methods. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper helps make it easier to evaluate how well people speak by using computers. Right now, the data we use to train these computer models is limited and unevenly spread out, which makes them not very reliable when they’re used in real-life situations. The authors came up with a new way to fix this problem by pre-training their model using some extra information and a special type of loss function. They tested it on a big dataset and found that it works really well, even when the data is different from what was trained on. |
Keywords
» Artificial intelligence » Loss function » Probability » Semi supervised
