no code implementations • 11 Nov 2022 • Yang Li, Canran Xu, Guodong Long, Tao Shen, Chongyang Tao, Jing Jiang
Basically, an instance-dependent soft prefix, derived from fact-counterfactual pairs in the label space, is leveraged to complement the language verbalizers in many-class classification.
no code implementations • WNUT (ACL) 2021 • Mengyi Gao, Canran Xu, Peng Shi
State-of-the-art approaches to spelling error correction problem include Transformer-based Seq2Seq models, which require large training sets and suffer from slow inference time; and sequence labeling models based on Transformer encoders like BERT, which involve token-level label space and therefore a large pre-defined vocabulary dictionary.
2 code implementations • 22 Nov 2019 • Canran Xu, Ming Wu
Learning representations for feature interactions to model user behaviors is critical for recommendation system and click-trough rate (CTR) predictions.
no code implementations • ACL 2019 • Canran Xu, Ruijiang Li
This new model learns knowledge graph embeddings that can capture relation compositions by nature.
Ranked #14 on Link Prediction on YAGO3-10
2 code implementations • 12 Nov 2018 • Xiang Wang, Dingxian Wang, Canran Xu, Xiangnan He, Yixin Cao, Tat-Seng Chua
Such connectivity not only reveals the semantics of entities and relations, but also helps to comprehend a user's interest.