- Level Expert
- المدة 48 ساعات hours
- الطبع بواسطة DeepLearning.AI
-
Offered by
عن
In the third course of Machine Learning Engineering for Production Specialization, you will build models for different serving environments; implement tools and techniques to effectively manage your modeling resources and best serve offline and online inference requests; and use analytics tools and performance metrics to address model fairness, explainability issues, and mitigate bottlenecks. Understanding machine learning and deep learning concepts is essential, but if you’re looking to build an effective AI career, you need production engineering capabilities as well. Machine learning engineering for production combines the foundational concepts of machine learning with the functional expertise of modern software development and engineering roles to help you develop production-ready skills. Week 1: Neural Architecture Search Week 2: Model Resource Management Techniques Week 3: High-Performance Modeling Week 4: Model Analysis Week 5: Interpretabilityالوحدات
A conversation with Andrew Ng, Robert Crowe and Laurence Moroney
1
Videos
- Course Overview
Hyperparameter tuning: searching for the best architecture
1
Assignment
- Hyperparameter Tuning and Neural Architecture Search
1
External Tool
- [IMPORTANT] Have questions, issues or ideas? Join our Community!
2
Videos
- Hyperparameter Tuning
- Keras Autotuner Demo
2
Readings
- Ungraded Lab - Intro to Keras Tuner
- Ungraded Lab - Hyperparameter Tuning and Model Training with TFX
AutoML
1
Assignment
- AutoML
2
External Tool
- A Tour of Qwiklabs and Google Cloud
- Classify Images of Clouds in the Cloud with AutoML Vision
7
Videos
- Intro to AutoML
- Understanding Search Spaces
- Search Strategies
- Measuring AutoML Efficacy
- AutoML on the Cloud
- Assignment Setup
- Week 1 Wrap Up
2
Readings
- Neural Architecture Search
- AutoML
Lecture Notes (Optional)
1
External Tool
- Lecture Notes W1
Dimensionality Reduction
1
Assignment
- Dimensionality Reduction
8
Videos
- Dimensionality Effect on Performance
- Curse of Dimensionality
- Curse of Dimensionality: an Example
- Manual Dimensionality Reduction
- Manual Dimensionality Reduction: Case Study
- Algorithmic Dimensionality Reduction
- Principal Components Analysis
- Other Techniques
3
Readings
- Ungraded Lab - Manual Feature Engineering
- Dimensionality Reduction Techniques
- Ungraded Lab - Algorithmic Dimensionality Reduction
Quantization and Pruning
1
Assignment
- Quantization and Pruning
5
Videos
- Mobile, IoT, and Similar Use Cases
- Benefits and Process of Quantization
- Post Training Quantization
- Quantization Aware Training
- Pruning
3
Readings
- Quantization
- Pruning
- Ungraded Lab - Quantization and Pruning
Lecture Notes (Optional)
1
External Tool
- Lecture Notes W2
High-Performance Modeling
1
Assignment
- High-Performance Modeling
1
External Tool
- Running Distributed TensorFlow using Vertex AI
3
Videos
- Distributed Training
- High-Performance Ingestion
- Training Large Models - The Rise of Giant Neural Nets and Parallelism
2
Readings
- Ungraded Lab - Distributed Strategies with TF and Keras
- High-Performance Modeling
Knowledge Distillation
1
Assignment
- Knowledge Distillation
3
Videos
- Teacher and Student Networks
- Knowledge Distillation Techniques
- Case Study - How to Distill Knowledge for a Q&A Task
2
Readings
- Ungraded Lab - Knowledge Distillation
- Knowledge Distillation
Lecture Notes (Optional)
1
External Tool
- Lecture Notes W3
Model Analysis Overview
1
Assignment
- Model Analysis
1
Videos
- Model Performance Analysis
1
Readings
- TensorBoard
Advanced Model Analysis and Debugging
1
Assignment
- Model Analysis and Debugging
1
External Tool
- Machine Learning with TensorFlow in Vertex AI
10
Videos
- Introduction to TensorFlow Model Analysis
- TFMA in Practice
- Model Debugging Overview
- Benchmark Models
- Sensitivity Analysis and Adversarial Attacks
- Adversarial Attack Demo
- Residual Analysis
- Model Remediation
- Fairness
- Measuring Fairness
7
Readings
- TensorFlow Model Analysis
- Ungraded Lab - Tensorflow Model Analysis
- Ungraded Lab - Model Analysis with TFX Evaluator
- Sensitivity Analysis and Adversarial Attacks
- Ungraded Lab - Fairness Indicators
- Model Remediation and Fairness
- [IMPORTANT] Information about the Week 4 assignment
Continuous Evaluation and Monitoring
1
Assignment
- Continuous Evaluation and Monitoring
1
Videos
- Continuous Evaluation and Monitoring
1
Readings
- Continuous Evaluation and Monitoring
Lecture Notes (Optional)
1
External Tool
- Lecture Notes W4
Explainable AI
1
Assignment
- Explainable AI
1
Videos
- Explainable AI
1
Readings
- Explainable AI
Interpretability
1
Assignment
- Interpretability
2
Videos
- Model Interpretation Methods
- Intrinsically Interpretable Models
1
Readings
- Interpretability
Understanding Model Predictions
1
Assignment
- Understanding Model Predictions
8
Videos
- Model Agnostic Methods
- Partial Dependence Plots
- Permutation Feature Importance
- Shapley Values
- SHapley Additive exPlanations (SHAP)
- Testing Concept Activation Vectors
- LIME
- AI Explanations
6
Readings
- Permutation Feature Importance
- Ungraded Lab - Permutation Feature Importance
- Ungraded Lab - Shapley Values
- Understanding Model Predictions
- TCAV and LIME
- AI Explanations
Lecture Notes (Optional)
1
External Tool
- Lecture Notes W5
Course Resources
1
Readings
- Course 3 Optional References
Acknowledgments
1
Readings
- Acknowledgements
Auto Summary
"Machine Learning Modeling Pipelines in Production" is an expert-level course designed for those in Data Science & AI, instructed by Coursera. The course focuses on developing models for various environments, managing modeling resources, and enhancing model fairness and explainability. Over five weeks, learners will explore neural architecture search, resource management, high-performance modeling, analysis, and interpretability. Ideal for those aiming to build production-ready AI skills, the course offers a comprehensive approach to machine learning engineering in production. Available via a Starter subscription, the course spans 2880 minutes.

Robert Crowe