Seminars from Adrien Bardet and Amira Barhoumi, PhD students at LIUM
Lieu: IC2, Salle des conseils
Intervenant: Adrien Bardet, Amira Barhoumi
A Study on Multilingual Transfer Learning in Neural Machine Translation : Finding the Balance Between Languages
Transfer learning is a solution to the problem of lack of data for a given language pair in machine translation. We will try to determine what are the best practices to achieve an effective transfer and thus maximize the results of our task where little data is available. Differences in data quantities between languages for multilingual systems are problematic for learning from balanced systems. This difference in quantity must be taken into account from the pre-processing stage to ensure optimal use of the system in its different languages. We will see the impact of this balance on machine translation results.
An Empirical Evaluation of Arabic-specific Embeddings for Sentiment Analysis
In this paper, we propose several specific embeddings in Arabic sentiment analysis (SA) framework. Indeed, Arabic is characterized by its agglutination and morphological richness contributing to great sparsity that could affect embedding quality. This work presents a rigorous study that compares different types of Arabic-specific embeddings. We evaluate them with 2 neural architectures: one based on convolutional neural network (CNN) and the other one based on Bidirectional Long Short-Term Memory Bi-LSTM. Experiments are done on the Large Arabic-Book Reviews corpus (LABR). Our best results boost previous published accuracy by 1.9%. Moreover, we experiment combination of our individual systems defining very confident decision, reaching an accuracy of 92.2% on 98.25% of LABR test dataset.