Artificial Intelligence and Deep Learning Models for Actuarial Applications

Lecture slides from UNSW’s ACTL3143 & ACTL5111 courses

Author

Dr Patrick Laub

These are the lecture slides from the “Artificial Intelligence and Deep Learning Models for Actuarial Applications” courses (coded ACTL3143 & ACTL5111) at UNSW.

Lecture Materials

Preliminaries Python

Week 1 Tabular Data

Week 2 Text

Weeks 7-10 Advanced Topics

Readings

The readings from the book will come mainly from Géron (2022), which is available through the UNSW Library’s access to O’Reilly Media texts. I’ll give references to the 3rd edition, but if you get your hands on a copy of the 2nd edition then that is also fine. Some readings will be from James et al. (2021) (or equivalently the the Python version James et al. (2023)) which is available online; you’ll need the 2nd edition for this (the deep learning chapter is not in the 1st edition). Note, if I say “read from A up to B”, that means to read A but stop at B (without reading it).

Week	Readings
0	Géron (2022): Chapter 1 “The Machine Learning Landscape”, Chapter 2 “End-to-End Machine Learning Project” (up to “Handling Text and Categorical Attributes”)
1	James et al. (2021): Sections 10.1 “Single Layer Neural Networks” & 10.2 “Multilayer Neural Networks” Géron (2022): Chapter 2 “End-to-End Machine Learning Project” (up to “Better Evaluation Using Cross-Validation”), Chapter 10 “Introduction to Artificial Neural Networks With Keras” (up to “Building Complex Models Using the Functional API”)
2	Géron (2022): Chapter 3 “Classification” (up to “Multilabel Classification”), Chapter 10 Section “Building Complex Models Using the Functional API”, Chapter 13 Section “Encoding Categorical Features Using Embeddings” James et al. (2021): Section 10.4 “Document Classification” Vajjala et al. (2020): Chapters 1 and 2 (up to “Modeling”), Chapter 3 “Text Representation”
3	James et al. (2021): Section 10.3 “Convolutional Neural Networks” Géron (2022): Chapter 14 “Deep Computer Vision Using Convolutional Neural Networks” (just skim through the specific historical architectures, like InceptionNet etc.)
4	James et al. (2021): Section 10.5 “Recurrent Neural Networks” Géron (2022): Chapter 15 “Processing Sequences Using RNNs and CNNs” Hyndman & Athanasopoulos (2018): Section 5.1-5.3 and 5.8
5	Schelldorfer & Wüthrich (2019)
7	Molnar (2020) Chapter 4 “Methods Overview” Charpentier (2024) Chapter 4 “Models: Interpretability, Accuracy, and Calibration”
8	Chollet (2021): Chapter 14 “Conclusions” .

Other useful resources include the Actuaries Institute’s Actuaries’ Analytical Cookbook and the Swiss Association of Actuaries’ Actuarial Data Science Tutorials.

Contributors

Eric Tian Dong
Michael Jacinto
Marcus Lautier
Sam Luo
Hang Nguyen
Melissa Renard
Gayani Thalagoda

Copyright

Patrick Laub

References

Charpentier, A. (2024). Insurance, biases, discrimination and fairness. Springer.

Chollet, F. (2021). Deep learning with Python. Simon and Schuster.

Géron, A. (2022). Hands-on machine learning with Scikit-Learn, Keras, and TensorFlow (3rd ed.). O’Reilly Media.

Hyndman, R. J., & Athanasopoulos, G. (2018). Forecasting: Principles and practice. OTexts.

James, G., Witten, D., Hastie, T., & Tibshirani, R. (2021). An Introduction to Statistical Learning: with Applications in R. Springer.

James, G., Witten, D., Hastie, T., Tibshirani, R., & Taylor, J. (2023). An Introduction to Statistical Learning: with Applications in Python. Springer.

Molnar, C. (2020). Interpretable machine learning.

Schelldorfer, J., & Wüthrich, M. V. (2019). Nesting classical actuarial models into neural networks. Available at SSRN 3320525.

Vajjala, S., Majumder, B., Gupta, A., & Surana, H. (2020). Practical natural language processing: a comprehensive guide to building real-world NLP systems. O’Reilly Media.