Human Adversarial Examples
Hal discusses innovative approaches to data labeling, emphasizing the use of human adversarial examples to test the fragility of NLP systems. By involving linguists as "breakers," the research reveals how minimal alterations can expose weaknesses in machine learning models. This collaboration between machine learning developers and domain experts opens new avenues for progress and creativity in the field.In this clip
From this podcast
The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence)
Language (Technology) Is Power: Exploring the Inherent Complexity of NLP Systems w/ Hal Daumé - #395
Related Questions
Is less labeled data needed for training machine learning models as discussed in the episode "Big Data Doesn't Exist" and the clip "Deep Learning Insights" featuring Ilya Sutskever (OpenAI Chief Scientist) - Building AGI, Alignment, Spies, Microsoft, & Enlightenment and Running Out of Reasoning Tokens?
Is less labeled data needed for training machine learning models as discussed in the episode Cognilytica and the clip Future of Data featuring Ilya Sutskever (OpenAI Chief Scientist) in the episode "Big Data Doesn't Exist" and the clip "Deep Learning Insights" - Building AGI, Alignment, Spies, Microsoft, & Enlightenment and Running Out of Reasoning Tokens?