Approaches to Fairness and Bias Mitigation in Natural Language Processing
summary
Covers fairness and bias in NLP models such as GPT and BERT, focusing on why bias arises from web-trained pre-trained language models and how that leads to amplification of human biases. Describes methods to detect and mitigate those biases and presents tools you can integrate into models to promote fairness.
AI-written summary · about
chapters (10)
- 0:00Start
- 14:33IAT - Implicit Association Test
- 15:20WEAT - Word Embedding
- 17:47Identifying Bias Direction
- 20:41Gender Swapping
- 21:44Word Embedding Bias Mitigation
- 23:18Data Manipulation Techniques
- 24:54Demographic Aware Training
- 27:56DATA MATTERS! AWARENESS IS VERY IMPORTANT. More data does not mean better data.
- 29:00Model Cards
related talks
50mWhat It Takes to Control Societal Bias in Natural Language Processing
38mSocial Implications of Bias in Machine Learning
01h 30mFairness Tales: How To Measure And Mitigate Unfair Bias in Machine Learning Models
42mTowards Inclusive AI: Understanding bias in ML models
41mUsing AI to Build Fair and Equitable Workplaces
40mResponsible AI Practices: Fairness in ML
37mKeynote: Ethical Machine Learning: Creating Fair Models in an Unjust World
42mFairness in Machine Learning: From Theory to Practice