In the rise of Big Data, not only the amount of data but also its diversity is ever increasing. Beyond 'traditional' data consisting of samples of a fixed number of interpretable variables, there is data such as free text, time series (financial transactions, power usage), audio (speech), images and video. These so called signals typically need to be processed such that meaningful variables can be extracted and structured prior to further usage in data analyses and machine learning applications.

This workshop is focused on making powerful data representations from signals for machine learning applications. In two consecutive parts, we will focus on feature engineering and feature learning in which we touch upon the following subjects, for all of which Python code is provided: - feature extraction using convolution and Fourier analysis - building bag-of-visual-word models from images - feature learning for dimensionality reduction - end-to-end training of deep convolutional networks - applying the feature -engineering and -learning techniques for time series, speech, and image classification

The workshop is suitable for data scientists with knowledge and/or experience in applying machine learning with python (e.g. numpy, scipy, scikit-learn, pandas).