Gravitational wave bursts, transient events lasting from milliseconds to minutes within the frequency band of current generation detectors, originate from a range of astrophysical phenomena, including core-collapse supernovae, neutron star glitches, and highly eccentric black hole mergers. Due to the complexity and diversity of these sources, their signal morphologies are often poorly modeled or completely unknown, making traditional matched-filter techniques ineffective for many target sources. More critically, detection methods must be sensitive to entirely unexpected phenomena, adopting an “eyes wide open” approach to enhance detection capabilities beyond known or predictable events. This chapter explores the integration of several machine learning techniques in the analysis of gravitational wave bursts, addressing the challenges posed by unmodeled and unknown signal morphologies and outlining the strategies developed to approach these signals with minimal assumptions.