Summary of Frequency Enhanced Pre-training For Cross-city Few-shot Traffic Forecasting, by Zhanyu Liu et al.
Frequency Enhanced Pre-training for Cross-city Few-shot Traffic Forecasting
by Zhanyu Liu, Jianrong Ding, Guanjie Zheng
First submitted to arxiv on: 3 Jun 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI)
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 The proposed Frequency Enhanced Pre-training Framework for Cross-city Few-shot Forecasting (FEPCross) tackles the challenge of developing cities collecting sufficient traffic data for accurate forecasting. FEPCross incorporates a novel Cross-Domain Spatial-Temporal Encoder that trains on self-supervised tasks, leveraging information from both time and frequency domains. The framework consists of pre-training and fine-tuning stages, with modules designed to enrich training samples and mitigate overfitting risk. Empirical evaluations on real-world traffic datasets demonstrate FEPCross’s exceptional efficacy, outperforming existing approaches. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Imagine trying to predict traffic patterns in a city without enough data. That’s what many developing cities face! Researchers came up with an idea called cross-city few-shot forecasting, where they use information from similar cities to help with the prediction. They found that by looking at the frequency of traffic patterns, they could make more accurate predictions. This new framework, FEPCross, uses a special type of computer model that looks at both time and frequency patterns to make these predictions. It also helps prevent the model from becoming too specialized in one city’s data. By testing it on real-world traffic data, they showed that FEPCross is much better than other methods! |
Keywords
» Artificial intelligence » Encoder » Few shot » Fine tuning » Overfitting » Self supervised
