Summary of Sequence Generation Modeling For Continuous Value Prediction, by Hongxu Ma et al.
Sequence Generation Modeling for Continuous Value Prediction
by Hongxu Ma, Kai Tian, Tao Zhang, Xuefeng Zhang, Chunjie Chen, Han Li, Jihong Guan, Shuigeng Zhou
First submitted to arxiv on: 28 Dec 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Information Retrieval (cs.IR)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary A novel Generative Regression (GR) framework is introduced for continuous value prediction (CVP), addressing challenges in short video recommendation. GR transforms numerical values into token sequences through structural discretization, preserving data fidelity while improving precision. This method employs curriculum learning with an embedding mixup strategy to bridge training-inference gaps. Experimental evaluations on five datasets demonstrate GR’s superiority over existing methods. Real-world A/B tests on Kuaishou validate its practical effectiveness. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary A new way is found to predict continuous values, like how much someone likes a video. This prediction helps recommend videos that people will enjoy. The old way of doing this wasn’t very good because it had trouble with big value ranges and didn’t handle data imbalance well. A better approach is needed. This new method uses ideas from language modeling to create a sequence of tokens, which keeps the original data’s information while making predictions more accurate. It even works well in real-life situations on a popular video platform called Kuaishou. |
Keywords
» Artificial intelligence » Curriculum learning » Embedding » Inference » Precision » Regression » Token
