SNK @ DANKMEMES: Leveraging Pretrained Embeddings for Multimodal Meme Detection (short paper)

English. In this paper, we describe and present the results of meme detection system, specifically developed and submitted for our participation to the first subtask of DANKMEMES (EVALITA 2020). We built simple classifiers, consisting in feed forward neural networks. They leverage existing pretrained embeddings, both for text and image representation. Our best system (SNK1) achieves good results in meme detection (F1 = 0.8473), ranking 2nd in the competition, at a distance of 0.0028 from the first classified. Italiano. In questo articolo, descriviamo e presentiamo i risultati di un sistema di individuazione dei meme, ideato e sviluppato per partecipare al primo subtask di DANKMEMES (EVALITA 2020). Abbiamo realizzato dei semplici classificatori, costituiti da una rete neurale feed-forward: essi sfruttano embedding preesistenti, per la rappresentazione numerica di testo e immagini. Il nostro miglior sistema (SNK1) raggiunge buoni risultati nell’individuazione dei meme (F1 = 0.8473) e si è classificato secondo nella competizione, ad una distanza di 0.0028 dal primo classificato. 1 System description 1.1 General approach and tools DANKMEMES (Miliani et al., 2020) is a task for meme recognition and hate speech/event identification in memes and is part of the EVALITA 2020 evaluation campaign (Basile et al., 2020). Copyright © 2020 for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0) For our participation to the first subtask of DANKEMES, we built simple classification models for meme detection. The main challenge is to effectively combine textual and image inputs. We tried to exploit the ability of pretrained embedding to represent the information present in text and images, paying a limited computational cost. To quickly build various prototypes of neural networks, we used Uber Ludwig framework (Molino et al., 2019): a toolbox built on top of TensorFlow, which facilitates and speeds up the training and testing of various models. We trained our models using Google Colaboratory, a hosted Jupyter notebook service, which provides free access to GPUs, with some resource and time limitations.