Summary of Overcoming Data Limitations in Internet Traffic Forecasting: Lstm Models with Transfer Learning and Wavelet Augmentation, by Sajal Saha et al.
Overcoming Data Limitations in Internet Traffic Forecasting: LSTM Models with Transfer Learning and Wavelet Augmentation
by Sajal Saha, Anwar Haque, Greg Sidebottom
First submitted to arxiv on: 20 Sep 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: None
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 paper explores effective internet traffic prediction in smaller ISP networks using transfer learning and data augmentation techniques with two LSTM-based models, LSTMSeq2Seq and LSTMSeq2SeqAtn. Initially trained on a comprehensive dataset provided by Juniper Networks, the models are applied to smaller datasets representing real internet traffic telemetry. The study reveals that while both models perform well in single-step predictions, multi-step forecasts are challenging, particularly in terms of long-term accuracy. In smaller datasets, LSTMSeq2Seq generally outperforms LSTMSeq2SeqAtn, indicating that higher model complexity does not necessarily translate to better performance. The models’ effectiveness varies across different network domains, reflecting the influence of distinct traffic characteristics. To address data scarcity, Discrete Wavelet Transform is used for data augmentation, leading to significant improvements in model performance, especially in shorter-term forecasts. The study includes an analysis of the models’ variability and consistency, with attention mechanisms in LSTMSeq2SeqAtn providing better short-term forecasting consistency but greater variability in longer forecasts. The results highlight the benefits and limitations of different modeling approaches in traffic prediction. Overall, this research underscores the importance of transfer learning and data augmentation in enhancing the accuracy of traffic prediction models, particularly in smaller ISP networks with limited data availability. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper explores how to predict internet traffic patterns better. It uses special kinds of neural networks called LSTM-based models, which are trained on a big dataset and then applied to smaller datasets. The study finds that these models work well for short-term predictions but struggle when predicting far into the future. When using less data, one model is slightly better than the other. The study also shows how adding noise to the data helps improve the predictions, especially for shorter time periods. Additionally, it looks at how consistent the models are in their predictions and finds that one model does a better job of making accurate short-term forecasts but gets worse as you look further ahead. Overall, this research highlights the importance of using techniques like transfer learning and data augmentation to make traffic prediction models more accurate, especially for smaller networks with limited data. |
Keywords
» Artificial intelligence » Attention » Data augmentation » Lstm » Transfer learning
