USEMP at MediaEval placing Task 2014

A. Popescu (1) , S. Papadopoulos (2) , I. Kompatsiaris (2)

1 DIASI - Département Intelligence Ambiante et Systèmes Interactifs
2 CERTH ITI - CERTH Information Technologies Institute

We describe the participation of the USEMP team in the Placing Task at MediaEval 2014. We submitted four textual runs which are inspired by CEA LIST's 2013 participation. Our entries are based on probabilistic place modeling but also exploit machine tag and/or user modeling. The best results were obtained when all these types of information are combined. The accuracy of automatic at 1km reaches 0:235 when using only training data provided by organizers and 0:441 with the use of external data.

Keywords: User Modeling; Training data

Domains: Computer Science [cs]

A. Popescu, S. Papadopoulos, I. Kompatsiaris. USEMP at MediaEval placing Task 2014. Multimedia Benchmark Workshop, MediaEval 2014, Oct 2014, Barcelona, Spain. ⟨cea-01844441⟩