Unified Interpretation of Softmax Cross-Entropy and Negative Sampling: With Case Study for Knowledge Graph Embedding
This work addresses a theoretical gap for researchers in knowledge graph embedding, enabling fairer comparisons between loss functions, though it is incremental as it builds on existing methods.
The paper tackled the lack of theoretical relationship between softmax cross-entropy and negative sampling loss functions in knowledge graph embedding, which hindered fair comparisons, and used Bregman divergence to unify their interpretation, with experimental validation on FB15k-237 and WN18RR datasets showing the findings hold in practice.
In knowledge graph embedding, the theoretical relationship between the softmax cross-entropy and negative sampling loss functions has not been investigated. This makes it difficult to fairly compare the results of the two different loss functions. We attempted to solve this problem by using the Bregman divergence to provide a unified interpretation of the softmax cross-entropy and negative sampling loss functions. Under this interpretation, we can derive theoretical findings for fair comparison. Experimental results on the FB15k-237 and WN18RR datasets show that the theoretical findings are valid in practical settings.