You're reading from Machine Learning Infrastructure and Best Practices for Software Engineers

Product type Book

Published in Jan 2024

Publisher Packt

ISBN-13 9781837634064

Pages 346 pages

Edition 1st Edition

Languages

Python

Concepts

Machine Learning

Author (1):

Miroslaw Staron

Table of Contents (24) Chapters

Preface

1. Part 1:Machine Learning Landscape in Software Engineering

2. Machine Learning Compared to Traditional Software

3. Elements of a Machine Learning System

4. Data in Software Systems – Text, Images, Code, and Their Annotations

5. Data Acquisition, Data Quality, and Noise

6. Quantifying and Improving Data Properties

7. Part 2: Data Acquisition and Management

8. Processing Data in Machine Learning Systems

9. Feature Engineering for Numerical and Image Data

10. Feature Engineering for Natural Language Data

11. Part 3: Design and Development of ML Systems

12. Types of Machine Learning Systems – Feature-Based and Raw Data-Based (Deep Learning)

13. Training and Evaluating Classical Machine Learning Systems and Neural Networks

14. Training and Evaluation of Advanced ML Algorithms – GPT and Autoencoders

15. Designing Machine Learning Pipelines (MLOps) and Their Testing

16. Designing and Implementing Large-Scale, Robust ML Software

17. Part 4: Ethical Aspects of Data Management and ML System Development

18. Ethics in Data Acquisition and Management

19. Ethics in Machine Learning Systems

20. Integrating ML Systems in Ecosystems

21. Summary and Where to Go Next

22. Index

Why subscribe?

23. Other Books You May Enjoy

Developing safety cages to prevent models from breaking the entire system

As GenAI systems such as MLMs and AEs create new content, there is a risk that they generate content that can either break the entire software system or become unethical.

Therefore, software engineers often use the concept of a safety cage to guard the model itself from inappropriate input and output. For an MLM such as RoBERTa, this can be a simple preprocessor that checks whether the content generated is problematic. Conceptually, this is illustrated in Figure 11.8:

Figure 11.8 – Safety-cage concept for MLMs

In the example of the wolfBERTa model, this can mean that we check whether the generated code does not contain cybersecurity vulnerabilities, which can potentially allow hackers to take over our system. This means that all programs generated by the wolfBERTa model should be checked using tools such as SonarQube or CodeSonar to check for cybersecurity vulnerabilities...