What You Need to Know Before
You Start
Starts 6 June 2025 09:17
Ends 6 June 2025
00
days
00
hours
00
minutes
00
seconds
All Models Are Wrong, Some Are Useful: Model Selection with Limited Labels
Discover MODEL SELECTOR, a framework that efficiently identifies the best pretrained classifier using minimal labeled data, reducing labeling costs by up to 94.15% across 1,500+ models on 16 datasets.
Scalable Parallel Computing Lab, SPCL @ ETH Zurich
via YouTube
Scalable Parallel Computing Lab, SPCL @ ETH Zurich
2484 Courses
25 minutes
Optional upgrade avallable
Not Specified
Progress at your own speed
Free Video
Optional upgrade avallable
Overview
Discover MODEL SELECTOR, a framework that efficiently identifies the best pretrained classifier using minimal labeled data, reducing labeling costs by up to 94.15% across 1,500+ models on 16 datasets.
Syllabus
- Introduction to Model Selection
- Overview of Pretrained Models
- Introduction to MODEL SELECTOR Framework
- Practical Guide to Using MODEL SELECTOR
- Strategies for Reducing Labeling Costs
- Case Studies
- Performance Evaluation
- Advanced Topics
- Ethics and Responsible Use of Pretrained Models
- Final Project: Applying MODEL SELECTOR to a Real-World Dataset
- Conclusion
Importance of Model Selection in Machine Learning
Challenges of Model Selection with Limited Labeled Data
Types of Pretrained Models
Advantages and Limitations
Concept and Objectives
Key Features and Benefits
Framework Architecture
Step-by-Step Implementation
Active Learning Approaches
Uncertainty Sampling Techniques
Model Selection on Image Classification Datasets
Model Selection on Text Classification Datasets
Metrics for Model Comparison
Analysis of Results and Cost Efficiency
Integrating MODEL SELECTOR with Popular Machine Learning Libraries
Customizing MODEL SELECTOR for Specific Use Cases
Bias and Fairness Considerations
Guidelines for Ethical Usage
Project Guidelines and Benchmarks
Presentation and Peer Review
Summary of Key Learnings
Future Trends in Model Selection and Label Efficiency
Subjects
Computer Science
