Suggest edit — Yuan 2021

Title

Name

Note

---
title: "Yuan 2021"
type: concept
related: [Adress Challenge, Loo]
source: https://www.jemoka.com/posts/kbhyuan_2021/
confidence: high
status: active
---

DOI: 10.3389/fcomp.2020.624488
One-Liner Used an ERNIE trained on transcripts for classification; inclusion of pause encoding made results better.
Novelty Instead of just looking at actual speech content, look at pauses specific as a feature engineering task \(89.6\%\) on the ADReSS Challenge dataset Notable Methods Applied FA with pause encoding with standard .cha semantics (short pauses, medium pauses, long pauses). Shoved all of this into an ERNIE.
Assay for performance was LOO
Key Figs Fig 1 This figure motivates the point that subjects with AD says oh and um more often; which prompted Table 1
Table 1 Subjects with AD says uh a lot more often; no significance level calculations but ok.
Figure 5 This figure is the result of a LOO study on the proposed model and presumably others before. X axis is the validation accuracy in question, Y is the density by which the score in X appears in an \(N=35\) LOO measurement.
This figure tells us that either way the ERNIE model is better than state of the art; furthermore, transcripts with pause encoding did better and did it better more of the time; that&rsquo;s where the 89.6% came from.
New Concepts Leave-One-Out cross validation Notes Glorious.