Summary of Online Data Augmentation For Forecasting with Deep Learning, by Vitor Cerqueira et al.
Online Data Augmentation for Forecasting with Deep Learning
by Vitor Cerqueira, Moisés Santos, Luis Roque, Yassine Baghoussi, Carlos Soares
First submitted to arxiv on: 25 Apr 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Machine Learning (stat.ML)
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 work introduces an online data augmentation framework for training neural networks on forecasting tasks with multiple univariate time series datasets. The existing methods rely on large enough training sample sizes, which may not always be available. To address this limitation, the authors employ synthetic data generation techniques that can be applied during the training process itself, rather than offline before model training. This approach maintains a balanced representation between real and synthetic data throughout training, eliminating the need to store large augmented datasets. The framework is validated using 3797 time series from 6 benchmark datasets, three neural architectures, and seven synthetic data generation techniques. |
Low | GrooveSquid.com (original content) | Low Difficulty Summary A new way of creating fake data for machine learning models has been developed. Right now, people usually create this fake data before training the model. But what if you could make it during training? This new method does just that! It creates fake data for each group of examples (called a batch) while training the model. This keeps the fake and real data balanced throughout the training process. The authors tested this method using many different datasets, models, and ways to create fake data. They found that this new approach works better than the old way and doesn’t require storing all the extra fake data. |
Keywords
» Artificial intelligence » Data augmentation » Machine learning » Synthetic data » Time series