It is argued that the analysis of the learner’s generated log files during interactions with a learning environment is necessary to produce interpretative views of their activities. The analysis of these log files, or traces, provides "knowledge" about the activity we call indicators. Our work is related to this research field. We are particularly interested in automatically identifying learners’ learning styles from learning indicators. This concept, used in several Educational Hypermedia Systems (EHS) as a criterion for adaptation and tracking, belongs to a set of behaviors and strategies in how to manage and organize information. In this paper, we validate our approach of auto-detection of student's learning styles based on their navigation behavior using machine-learning classifiers.