Affinity Weighted Embedding
This addresses the limitation of linear models in leveraging large datasets for practitioners in ranking and recommendation, though it appears incremental as it builds on existing embedding methods.
The authors tackled the underfitting problem of linear embedding models like Wsabie and PSI in ranking, recommendation, and annotation tasks by proposing a new class of models that iteratively reweight features and labels, showing improved performance.
Supervised (linear) embedding models like Wsabie and PSI have proven successful at ranking, recommendation and annotation tasks. However, despite being scalable to large datasets they do not take full advantage of the extra data due to their linear nature, and typically underfit. We propose a new class of models which aim to provide improved performance while retaining many of the benefits of the existing class of embedding models. Our new approach works by iteratively learning a linear embedding model where the next iteration's features and labels are reweighted as a function of the previous iteration. We describe several variants of the family, and give some initial results.