Classification of the data using the extracted features

The aim of the workshop is to introduce students into application of machine learning methods to classify EUSO-SPB1 data. The presentation includes a brief introduction into the topic and examples of expected results. Students can use data created by their own feature extraction procedure or a dataset provided for this workshop. In this exercise, students can experiment with a variety of classification methods and afterwards compare their results.

The Air Shower Pattern Recognition summer school consists of the following five workshops:

1. Introduction to the JEM-EUSO and running air shower event simulations
2. Reading the detector data
3. Implementation of a feature extraction method
4. Classification of the data through the extracted features
5. Classification of the data using a convolutional neural network

Slides

Practical activity

1. Loading and preparation of the data for analysis – train-test split.
2. Feature selection and training of decision tree-based classifiers
3. Evaluation of the classification accuracy.
4. Application of a T-SNE method, visualization of the embedding