CLAIOct 20, 2021

Distributionally Robust Classifiers in Sentiment Analysis

arXiv:2110.10372v1
Originality Incremental advance
AI Analysis

This work addresses robustness in sentiment analysis for applications facing data distribution changes, but it is incremental as it builds on existing BERT and DRO methods.

The authors tackled sentiment classification under distributional shifts by integrating BERT with Distributionally Robust Classifiers (DRO), achieving improved performance on a test set with a shift from IMDb to Rotten Tomatoes datasets.

In this paper, we propose sentiment classification models based on BERT integrated with DRO (Distributionally Robust Classifiers) to improve model performance on datasets with distributional shifts. We added 2-Layer Bi-LSTM, projection layer (onto simplex or Lp ball), and linear layer on top of BERT to achieve distributionally robustness. We considered one form of distributional shift (from IMDb dataset to Rotten Tomatoes dataset). We have confirmed through experiments that our DRO model does improve performance on our test set with distributional shift from the training set.

Code Implementations1 repo
Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes