Soft Token Matching for Interpretable Low-Resource Classification

Federico Errica, Fabrizio Silvestri, Bora Edizel, Sebastian Riedel, Ludovic Denoyer, Vassilis Plachouras

Abstract:

We propose a model to tackle classification tasks in the presence of very little training data. To this aim, we introduce a novel matching mechanism to focus on elements of the input by using vectors that represent semantically meaningful concepts for the task at hand.
By leveraging highlighted portions of the training data, a simple, yet effective, error boosting technique guides the learning process. In practice, it increases the error associated to relevant parts of the input by a given factor. Results on text classification tasks confirm the benefits of the proposed approach in both balanced and unbalanced cases, thus being of practical use when labeling new examples is expensive. In addition, the model is interpretable, as it allows for human inspection of the learned weights.

Code: https://www.dropbox.com/s/5ygc44k56954qtf/PARCUS_CODE.zip?dl=0

Keywords: low-resource classification, semantic matching, error boosting, text classification, natural language processing

TL;DR: We propose a model to efficiently tackle low resource classification, which leverages additional information in the input to guide the learning process.