Video Title:
Clean Machine Learning Code: Practical Software Engineering Principles for ML Craftsmanship | Moussa Taifi Ph.D | PyData New York 2019
Contents:
00:00 Introduction
00:54 Smelly Storytime pt/1/2
01:48 Smelly Storytime pt/2/2
03:22 How can CMLC principles help?
03:42 Why care about CMLC?
05:02 How is the ML workforce-2019?
06:22 Original "Clean Code" books
06:32 What are some benefits of CMLC?
07:32 Clean Code Cheat Sheet
08:42 Practical Principles for Clean ML Code
09:11 Loose Coupling
10:22 High Cohesion
10:52 Change is Local
11:32 It is Easy to Remove
12:01 Mind-sized Components
14:32 Useful Design Principles
15:42 SOLID Design Principles
17:35 Single Responsibility Principle SRP
19:15 Open-Closed Principle OCP
20:15 Liskov Substitution Principle LSP
21:35 Interface Segregation Principle ISP
21:55 Dependency-inversion Principle DIP
23:45 Impact of TDML: Test Driven Machine Learning
24:36 How violating these principles leads to ML Tech Debt?
25:15 Entanglement and Glue Code
26:25 Hidden Feedback Loops
28:25 Undeclared Consumers
29:45 Pipeline Jungles
32:55 ML Tech Debt vs. Clean Code Principles
34:25 The Future of ML Engineering Without Self-Regulation
36:35 Thanks - Summary Slide
37:10 How do you measure your quality Question
40:10 Undeclared Data Customers Quest ion
40:23 All Design is Deduplication
41:47 Refactoring Jupyter Notebooks with IDEs Question
Video link: https://www.youtube.com/watch?v=PEjTAJHxYPM
Video Title: Clean Machine Learning Code: Practical Software Engineering Principles for ML Craftsmanship | Moussa Taifi Ph.D | PyData New York 2019
Contents: 00:00 Introduction 00:54 Smelly Storytime pt/1/2 01:48 Smelly Storytime pt/2/2 03:22 How can CMLC principles help? 03:42 Why care about CMLC? 05:02 How is the ML workforce-2019? 06:22 Original "Clean Code" books 06:32 What are some benefits of CMLC? 07:32 Clean Code Cheat Sheet 08:42 Practical Principles for Clean ML Code 09:11 Loose Coupling 10:22 High Cohesion 10:52 Change is Local 11:32 It is Easy to Remove 12:01 Mind-sized Components 14:32 Useful Design Principles 15:42 SOLID Design Principles 17:35 Single Responsibility Principle SRP 19:15 Open-Closed Principle OCP 20:15 Liskov Substitution Principle LSP 21:35 Interface Segregation Principle ISP 21:55 Dependency-inversion Principle DIP 23:45 Impact of TDML: Test Driven Machine Learning 24:36 How violating these principles leads to ML Tech Debt? 25:15 Entanglement and Glue Code 26:25 Hidden Feedback Loops 28:25 Undeclared Consumers 29:45 Pipeline Jungles 32:55 ML Tech Debt vs. Clean Code Principles 34:25 The Future of ML Engineering Without Self-Regulation 36:35 Thanks - Summary Slide 37:10 How do you measure your quality Question 40:10 Undeclared Data Customers Quest ion 40:23 All Design is Deduplication 41:47 Refactoring Jupyter Notebooks with IDEs Question
Resources: Abstract: https://pydata.org/nyc2019/schedule/presentation/20/clean-machine-learning-code-practical-software-engineering-principles-for-ml-craftsmanship/ Book: https://leanpub.com/cleanmachinelearningcode Course: https://www.udemy.com/course/clean-machine-learning-code