On the scaling of polynomial features for representation matching
This work addresses a specific technique for improving representation matching in neural models, but it is incremental as it focuses on optimizing scaling for polynomial features without introducing a new paradigm.
The paper tackled the problem of scaling polynomial features for representation matching in neural models, specifically using natural language inference as an example, and found that scaling degree 2 features reduced classification error by 5% in the best models.
In many neural models, new features as polynomial functions of existing ones are used to augment representations. Using the natural language inference task as an example, we investigate the use of scaled polynomials of degree 2 and above as matching features. We find that scaling degree 2 features has the highest impact on performance, reducing classification error by 5% in the best models.