论文阅读 《Progressive Self-Supervised Attention Learning for Aspect-Level Sentiment Analysis》
2020-04-20 21:50
507 查看
这篇文章也是针对aspect-level SA的一篇文章，其实准确的说更像一篇针对注意力机制的文章，提出了SA模型目前注意力机制的缺陷：过度注意高频词，忽略了低频次，本文用了两个技巧：第一个是迭代的mask注意力权重最大的词，第二个是加入了一个正则化项
内容来自用户分享和网络整理，不保证内容的准确性，如有侵权内容，可联系管理员处理 
相关文章推荐
- 论文阅读《ActiveStereoNet:End-to-End Self-Supervised Learning for Active Stereo Systems》
- 【CV论文阅读】Unsupervised deep embedding for clustering analysis
- 论文阅读：Attention-based Dropout Layer for Weakly Supervised Object Localization
- 情感分析论文阅读之《Aspect Level Sentiment Classification with Deep Memory Network》
- [论文阅读]Attention U-Net: Learning Where to Look for the Pancreas
- 论文阅读笔记（1）：A Model of Saliency-Based Visual Attention for Rapid Scene Analysis
- 论文阅读：Two-Phase Learning for Weakly Supervised Object Localization
- 阅读图像显著性检测论文一：A Model of Saliency-Based Visual Attention for Rapid Scene Analysis
- 论文阅读笔记 | (IJCAI 2018 Oral) Collaborative Learning for Weakly Supervised Object Detection
- ResNet: Deep Residual Learning for Image Recognition 论文阅读
- 论文阅读：《LIP: Self-supervised Structure-sensitive Learning and A New Benchmark for Human Parsing》
- 行为识别论文阅读（1）——Deep Progressive Reinforcement Learning for Skeleton-based Action Recognition
- 论文阅读理解 - Multi-Context Attention for Human Pose Estimation
- 论文阅读：Compositional Learning for Human Object Interaction
- 论文阅读Soft Proposal Networks for weekly-supervised object localization
- 图像显著性论文（一）—A Model of saliency Based Visual Attention for Rapid Scene Analysis
- Aspect Specific Sentiment Analysis using Hierarchical Deep Learning (Lakkaraju, 2014)
- （MTL）论文阅读CVPR2019：end-to-end multi-task learning with attention
- 论文阅读理解 - Deep Learning of Binary Hash Codes for Fast Image Retrieval
- 论文阅读学习 - ResNet - Deep Residual Learning for Image Recognition