PyData London 2023

A dive into Hyperparameter Optimization
06-02, 09:00–10:30 (Europe/London), Salisbury

The tutorial aims to introduce the audience to the power of Hyperparameter Optimization. It will help them learn; how using simple python libraries one can make a huge difference in their ML model behavior.

We start with understanding the importance of hyperparameters, and the different distributions they are selected from. We then review some basic methods of optimizing hyperparameters, moving on to distributed methods and then to bayesian optimization methods. We'll use these algorithms hands-on, and play around with search spaces. We'll try out packages like Hyperopt, Dask, Optuna, to tune hyperparameters.

This tutorial will help beginner-level ML practitioners and working professionals use these methods in their applied ML tasks. They will be able to enhance the model quality and tune hyperparameters for bulky experiments more effectively.

Prior Knowledge Expected - Basic Python, a very basic understanding of Machine learning.\
Good to have - worked with libraries like scikit-learn(just knowing should be enough)

I have seen most ML practitioners tune their hyperparameters with algorithms like Grid Search, and Random Search. But after this tutorial, they'll be able to use more advanced algorithms, saving time while increasing model effectiveness.

We'll go through three segments in this workshop; Slide Presentation, a hands-on session on Jupyter Notebook followed by a QA session.

Segment 1 - slides link(please note that you can scroll down in slides):\
In these slides, we'll go through:\
* Hyperparameters
* Types of Distributions(to create search spaces)
* Grid Search and Random Search
* Distributed Optimization using Dask
* Bayesian Optimization
* Sequential Model-based Global Optimization(with TPE)
* Hyperopt- library overview\
We'll look here how these algorithms actually work, and how we decide search space and distributions for hyperparameters.

Segment 2 - code link:\
(45 mins)\
In this segment, we'll one by one try hands-on code for all the above optimization algorithms.

Segment 3:\
(5-10 mins)\
A small QA session to answer all the doubts from the audience.

Make sure you have Jupyter lab/ Jupyter Notebook installed, and I'll take care of the rest.

Alternatively you can use this link to open notebook in colab, some features of notebook won't work as aspected, but I'll show you while presenting my screen anyways.

Prior Knowledge Expected

Previous knowledge expected

Tanay Agrawal is Deep Learning Engineer, currently working with Curl HG. He specializes in Computer Vision and Deep Learning. He has extensively worked on Hyperparameter Optimization. He has published a book on the same; "Hyperparameter Optimization in Machine Learning" with Apress.