横井 祥

Norm of word embedding encodes information gain 

2022年12月19日  

Improving word mover's distance by leveraging self-attention matrix 

CoRR abs/2211.06229 2022年  

Subspace-based Set Operations on a Pre-trained Word Embedding Space 

CoRR abs/2210.13034 2022年  

Instance-Based Neural Dependency Parsing 

Transactions of the Association for Computational Linguistics 9 1493-1507 2021年12月17日  査読有り

Computationally Efficient Wasserstein Loss for Structured Labels 

CoRR abs/2103.00899 2021年  

訓練事例の影響の軽量な推定 

自然言語処理 28(2) 573-597 2021年  査読有り

Incorporating Residual and Normalization Layers into Analysis of Masked Language Models 

Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing 4547-4568 2021年  査読有り

Computationally Efficient Wasserstein Loss for Structured Labels 

Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics: Student Research Workshop 1-7 2021年  査読有り

Evaluation of Similarity-based Explanations 

The Ninth International Conference on Learning Representations (ICLR) 2021年  査読有り

Modeling Event Salience in Narratives via Barthes' Cardinal Functions 

CoRR abs/2011.01785 2020年  

Word Rotator's Distance: Decomposing Vectors Gives Better Representations 

CoRR abs/2004.15003 2020年  

Utterance Pair Scoring for Noisy Dialogue Data Filtering 

CoRR abs/2004.14008 2020年  

Efficient Estimation of Influence of a Training Instance 

Proceedings of SustaiNLP: Workshop on Simple and Efficient Natural Language Processing 2020年  査読有り

Attention is Not Only a Weight: Analyzing Transformers with Vector Norms 

Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP) 7057-7075 2020年  査読有り