paper-with-me

홈 › Papers

Unified Interpretation of Softmax Cross-Entropy and Negative Sampling: With Case Study for Knowledge Graph Embedding

2021-06-14 · ACL 2021 5 · Hidetaka Kamigaito, Katsuhiko Hayashi

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.

📄 PDF Abstract BibTeX arXiv:2106.07250

Code (1)

kamigaito/acl2021kge 공식 구현 pytorch

Tasks

Graph EmbeddingKnowledge Graph EmbeddingLink Predictionvalid

Methods 이 논문이 사용한 방법론

Softmax The Softmax output function transforms a previous layer's output into a vector of probabilities. It is commonly used for multiclass classification. Given an input vector $x$…

Similar Papers 제목 키워드 기반

UniFace: Unified Cross-Entropy Loss for Deep Face Recognition

2023-01-01 · ICCV 2023 1 · Jiancan Zhou, Xi Jia, Qiufu Li, Linlin Shen 외

As a widely used loss function in deep face recognition, the softmax loss cannot guarantee that the minimum positive sample-to-class similarity is larger than the maximum negative sample-to-class similarity. As a res…

Face RecognitionTAR

Similarity search generalisation in contrastive learning with InfoNCE loss

2026-07-10 · Nick Whiteley arxiv

Similarity search is a primary application of embedding models trained by contrastive learning. For one of the most popular contrastive learning loss functions, InfoNCE, we show that the population risk with $k$ negative…

Contrastive Learning

SimCE: Simplifying Cross-Entropy Loss for Collaborative Filtering

2024-06-23 · Xiaodong Yang, Huiyuan Chen, Yuchen Yan, Yuxin Tang 외

The learning objective is integral to collaborative filtering systems, where the Bayesian Personalized Ranking (BPR) loss is widely used for learning informative backbones. However, BPR often experiences slow convergence…

Collaborative FilteringRecommendation Systems

Revisiting Softmax Masking: Stop Gradient for Enhancing Stability in Replay-based Continual Learning

2023-09-26 · Hoyong Kim, Minchan Kwon, Kangil Kim

In replay-based methods for continual learning, replaying input samples in episodic memory has shown its effectiveness in alleviating catastrophic forgetting. However, the potential key factor of cross-entropy loss with …

Continual LearningIncremental Learning

On Expected Accuracy

2019-05-01 · Ozan İrsoy

We empirically investigate the (negative) expected accuracy as an alternative loss function to cross entropy (negative log likelihood) for classification tasks. Coupled with softmax activation, it has small derivatives o…

ClassificationGeneral Classificationimage-classificationImage Classification+1