Learn Before
Choosing a Machine Learning Strategy with Limited Data
A medical research team has a small, highly specialized dataset of 500 labeled images for a new diagnostic task. They are considering two strategies: (1) Training a new model from scratch using only their 500 images, or (2) Adapting a large, general-purpose model (pre-trained on millions of general images) using their 500 images. Which strategy is more advisable, and why? Justify your choice based on the principle of how effectively a model can learn from a limited number of examples.
0
1
Tags
Ch.4 Alignment - Foundations of Large Language Models
Foundations of Large Language Models
Foundations of Large Language Models Course
Computing Sciences
Evaluation in Bloom's Taxonomy
Cognitive Psychology
Psychology
Social Science
Empirical Science
Science
Related
Instruction Fine-Tuning as a Sample-Efficient Method
Application of Sample Efficiency in Advanced Learning Techniques
Two teams are developing a machine learning model to classify customer support tickets. Team A adapts a large, general-purpose model using 1,000 labeled tickets and achieves 92% accuracy. Team B trains a new model from scratch using 100,000 labeled tickets and also achieves 92% accuracy. Based on this information, which statement correctly analyzes the two approaches?
Choosing a Machine Learning Strategy with Limited Data
Evaluating a Development Strategy for an AI Chatbot