Stanford Practical Machine Learning-课程介绍

这门课程主要关于，机器学习在工业界的运用。教授一个数据科学家，将机器学习用到工业界的时候后，在不同的阶段所遇到了一些比较重要的技术细节。

Industrial ML Applications

• Manufacturing: Predictive maintenance, quality control
• Retail: Recommendation, chatbox, demand forecasting
• Healthcare: Alerts from real-time patient data, disease identification
• Finance: Fraud detection, application processing
• Automobile: Breakdown prediction, self-driving
• House Price Prediction: The goal is to predict the bid price for the winning buyer

ML Workflow

• Problem formulation
• A loop:
  • Collect & process data
  • Train & Tune models
  • Deploy models
  • Monitor

ML Challenges

• Formulate problem: focus on the most impactful industrial problems
• Data: high-quality data is scarce, privacy issues
• Train models: ML models are more and more complex, data-hungry, expensive
• Deploy models: heavy computation is not suitable for real-time inference
• Monitor: data distributions shifts, fairness issues

Roles

• Domain experts: have business insights, know what data is important and where to find it, identify the real impact of a ML model

• Data scientists: full stack on data mining, model training and deployment

• ML experts: customize SOTA ML models

• SDE: develop/maintain data pipelines, model training and serving pipelines

• Skill Improvement:

SDE和领域专家也会慢慢向数据科学家靠拢，数据科学家慢慢会成为机器学习专家。

• How data scientists spent their time (source: Anaconda survey 2020)

Course Topics

• Techniques a data scientist needs but often not taught in university ML/stats/programming courses
  • Data
    • Collect/ preprocess data
    • Covariate/ concepts/label shifts
    • Data beyond IID
  • Train
    • Model validation/combinations/tuning
    • Transfer learning
    • Multi-modality
  • Deploy
    • Model deployment
    • Distillation
  • Monitor
    • Fairness
    • Explainability

References

1. 课件
2. 课程官网
3. Slides