Generating Long Financial Report using Conditional Variational Autoencoders with Knowledge Distillation

• DOI: 10.1609/aaai.v35i18.17936
• Publication Date: 2022-09-08T20:38:19Z
• Authors: Yunpeng Ren; Ziao Wang; Yiyuan Wang; Xiaofeng Zhang
• Keywords: FOS: Computer and information sciences; Computer Science - Machine Learning; 01 natural sciences; Machine Learning (cs.LG); 0105 earth and related environmental sciences

Automatically generating financial report from a piece of news is quite challenging task. Apparently, the difficulty this task lies in lack sufficient background knowledge to effectively generate long report. To address issue, paper proposes conditional variational autoencoders (CVAE) based approach which distills external corpus news-report data. Experimental results demonstrate that proposed could achieve SOTA performance.