Active learning and weak supervision in NLP projects
summary
Describes using active learning (including text embeddings for task-specific sampling) to reduce required labeled examples for NLP, explains weak supervision by writing simple labeling rules to generate large training sets without manual labels, and outlines combining active learning and weak supervision to improve model metrics.
AI-written summary · about
related talks
51mEfficient Deep Learning with Humans in the Loop
39mActive learning in the interactive python environment
23mLearning with Limited Labels & Weak Supervision
37mActive Teaching, Human Learning
30mActive Learning and Human-in-the-Loop Machine Learning
30mEfficient data labelling with weak supervision
33mWeak supervision: a new paradigm for unreliable data
55mKeynote - Natural Language Processing: Challenges and Next Frontiers