Mandarin Chinese Bai Li Advised by Frank Rudzicz Li B ., Hsu Y-T., - - PowerPoint PPT Presentation

▶
mandarin chinese
SMART_READER_LITE
LIVE PREVIEW

Mandarin Chinese Bai Li Advised by Frank Rudzicz Li B ., Hsu Y-T., - - PowerPoint PPT Presentation

Feb 21, 2024 386 likes •713 views

Automatic Detection of Dementia in Mandarin Chinese Bai Li Advised by Frank Rudzicz Li B ., Hsu Y-T., Rudzicz F. Detecting dementia in Mandarin Chinese using transfer learning from a parallel corpus. To appear at NAACL 2019. Alzheimers

slide-1
SLIDE 1

Automatic Detection of Dementia in Mandarin Chinese

Bai Li

Advised by Frank Rudzicz

Li B., Hsu Y-T., Rudzicz F. “Detecting dementia in Mandarin Chinese using transfer learning from a parallel corpus”. To appear at NAACL 2019.

slide-2
SLIDE 2

Alzheimer’s Disease (AD) and Dementia

  • Neurodegenerative disease
  • 5.7 million patients in the USA, 50 million worldwide
  • Symptoms:
  • Early: forgetfulness, language impairment
  • Late: loss of motor control, death
  • One of the most costly diseases
  • No cure is known
  • For this presentation, Alzheimer’s disease ≈ Dementia

2

slide-3
SLIDE 3

Why detect Alzheimer’s disease?

  • Early treatment
  • No known drugs to slow down progression of AD
  • But can reduce symptoms!
  • Clinical trials
  • Current treatments may be ineffective because started too late!

3

slide-4
SLIDE 4

Detecting AD

  • Many tests: MRI, PET scan
  • Cognitive tests
  • Category naming
  • Picture naming
  • Picture description
  • Cognitive tests: cheap, non-intrusive,

screening mechanism

4

slide-5
SLIDE 5

Category Fluency

  • Name as many {animals, fruits, colours} as possible in 60 seconds

5

slide-6
SLIDE 6

Picture Description

  • Describe this picture in as much detail as possible

6

slide-7
SLIDE 7

Linguistic impairment of AD

  • People with dementia use language

differently!

  • Word finding difficulties
  • “the boy is standing on the chair”
  • More pronouns / adverbial constructions
  • “he’s reaching up there”
  • Acoustic abnormality
  • Higher pause rate, slower speech
  • Less complex sentences

7

slide-8
SLIDE 8

Feature extraction for AD detection

Automated tools to extract relevant features

  • Length of narration
  • Vocabulary diversity
  • Type-token Ratio:

𝑣𝑜𝑗𝑟𝑣𝑓 𝑥𝑝𝑠𝑒𝑡 𝑢𝑝𝑢𝑏𝑚 𝑥𝑝𝑠𝑒𝑡

  • Frequency metrics in corpus

8

slide-9
SLIDE 9

Syntactic features

  • Part-of-speech tag counts (e.g: #adj,

#noun, #pronoun/#noun)

  • Constituency parse tree
  • Max, mean, median heights
  • Production rule counts
  • Length of clauses, dependent clauses,

coordinate phrases

  • Dependency parse tree
  • Mean, median, max dependency distance

9

slide-10
SLIDE 10

Machine learning to detect dementia

  • Fraser (2016) extracts over 400 features and achieves 81%

classification accuracy using logistic regression

  • Fraser, Kathleen C., Jed A. Meltzer, and Frank Rudzicz. "Linguistic features identify Alzheimer’s

disease in narrative speech." Journal of Alzheimer's Disease 49.2 (2016): 407-422.

10

slide-11
SLIDE 11

DementiaBank

  • Collected between 1983 to 1988 at University of Pittsburgh
  • 551 cookie theft narrations (241 healthy, 310 dementia)
  • Mini Mental State Exam (MMSE), scored out of 30
  • Other tasks
  • Demographic information, diagnosis

11

slide-12
SLIDE 12

Mandarin Dataset: Lu Corpus

  • 49 speakers of Taiwanese Mandarin
  • Several tasks for each speaker
  • Cookie theft picture description
  • Category Fluency (animals, fruits, colours, places in Taiwan)
  • Picture Naming (30 items)
  • Transcripts of the picture description available
  • Diagnostic information unknown

12

slide-13
SLIDE 13

Dementia Score using PCA

  • Derive a proxy score for dementia

13

slide-14
SLIDE 14

How to detect dementia in Chinese

  • Q: Why not just do the same thing that we did with English?
  • A: Not enough data
  • Solution: Need to combine datasets somehow, across

different languages

  • Use transfer learning!

14

slide-15
SLIDE 15

Some Domain Adaptation Methods

15

  • Large corpus in domain S, small corpus in domain T
  • Want accurate model for domain T
  • Existing methods require same features in S and T

Daume III, Hal. "Frustratingly Easy Domain Adaptation." Proceedings of the 45th Annual Meeting of the Association of Computational Linguistics. 2007.

slide-16
SLIDE 16

Cross-language features: Difficulties

she's standing on a chair 她站在一个椅子上

she/PRP 's/VBZ standing/VBG

  • n/IN a/DT chair/NN

她/PN 站/VV 在/P 一/CD 个/M 椅子/NN 上/LC

????

she-stand-at-one-CL-chair-on

16

slide-17
SLIDE 17

Cross-language features: Difficulties

  • Experiments: poor accuracy with universal cross-language

features

  • Model needs to learn not only to detect dementia
  • It also needs to learn how features correspond across languages
  • We only have n=49 samples!

17

slide-18
SLIDE 18

Idea: Extract features separately, learn correspondences using out-of-domain data!

18

slide-19
SLIDE 19

Movie subtitles!

  • OpenSubtitles corpus, containing aligned subtitles in 62 languages

19

slide-20
SLIDE 20

Baselines

  • 1. Unilingual: train model using Mandarin data only, evaluate

using cross-validation

  • 2. Google Translate: translate Mandarin narration into

English, then run the English classifier

20

slide-21
SLIDE 21

Evaluation: Spearman’s correlation between model’s output and dementia score

21

slide-22
SLIDE 22

Proposed Model (Learning Feature Correspondence)

  • 1. Extract feature vector 𝑦 in Chinese
  • 2. Extract feature vector 𝑧 in English, independently
  • 3. Learn mapping function 𝑔: 𝑦 → 𝑧 using OpenSubtitles

movie dialogue corpus This is a multi-output regression problem

22

slide-23
SLIDE 23

Proposed Model (Learning Feature Correspondence)

23

Unsupervised – only English dementia data used during training!

slide-24
SLIDE 24

Independent Linear Regressions

  • For each target feature, train a separate linear regression
  • Use ElasticNet regularization, independent hyperparameter search

24

… …

num_characters pronoun_count num_words noun_verb_ratio

Chinese English

slide-25
SLIDE 25

Reduced Rank Regression

  • Problem: not taking advantage of relationship between outputs
  • Solution:
  • Note: equivalent to linear neural network with hidden layer of size 𝑆

25

… …

num_characters pronoun_count num_words noun_verb_ratio

Chinese English

slide-26
SLIDE 26

Joint Feature Selection

  • Problem: some features are noisy or impossible to reconstruct
  • Solution: order by 𝑆2, use only the top 𝐿 features

26

… …

num_characters pronoun_count num_words noun_verb_ratio

𝑆2 = 0.8 𝑆2 = 0.1

Chinese English

slide-27
SLIDE 27

Results

  • Initial model not very good
  • Reduced rank regression also not effective
  • Joint feature selection beats baselines

27

𝑞 = 0.06

slide-28
SLIDE 28

Results: Number of features

28

Accuracy of English classifier using K features Performance of whole model

slide-29
SLIDE 29

29

slide-30
SLIDE 30

Ablation study

About 1000-2000 parallel sentences needed

30

slide-31
SLIDE 31

Summary

First use of NLP to detect dementia in Mandarin Chinese

  • 1. Extracted lexicosyntactic features in English and Chinese
  • 2. Used out-of-domain corpus to learn correspondence model
  • 3. Combined with English dementia classifier

31

slide-32
SLIDE 32

Future Work

  • Need for human transcripts
  • Incorporate speech data
  • Apply to other languages (French, Korean)
  • Collect quality data in multiple languages

32