* La présentation sera donnée en français.

Membership Query Synthesis (MQS) is an active learning paradigm in which one labels generated artificial examples instead of genuine ones to extend a dataset. Despite prodigious advances in the power of generative models, an essential component of MQS, the field stays severely under-studied, especially in the textual domain. In this paper, we show that the previous SOTA on textual MQS, which was tested on a curated dataset of short sentences, performs poorly when used on a real dataset. We propose and report better results than randomly selecting unlabelled data with random selection in the latent space of a VAE coupled with a simple set of filtering mechanisms. This provides an improvement of 31.1% over the previous MQS SOTA on the SST-2 dataset, and of 2.7% over random active learning. To the best of our knowledge, this is the first time MQS is reported to work on a textual task with no constraints on the size of the sentences.

Enregistrement : https://drive.google.com/file/d/1G0DAzf4hSURE9D0r0x_tN6jaA3IZOzHK/view?usp=sharing