Summary of Tuning Frequency Bias Of State Space Models, by Annan Yu et al.
Tuning Frequency Bias of State Space Models
by Annan Yu, Dongwei Lyu, Soon Hoe Lim, Michael W. Mahoney, N. Benjamin Erichson
First submitted to arxiv on: 2 Oct 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 paper explores state space models (SSMs), which use linear time-invariant systems to learn sequences with long-range dependencies. It finds that SSMs exhibit an implicit bias towards capturing low-frequency components, aligning with frequency bias in deep learning model training. The authors show that the initialization of an SSM assigns it a frequency bias and propose two mechanisms to tune this bias: scaling the initialization or applying a Sobolev-norm-based filter. Empirically, they demonstrate improved performance on long-range sequence learning tasks, achieving 88.26% accuracy on the Long-Range Arena (LRA) benchmark. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper talks about special kinds of computer models called state space models. These models are good at understanding things that happen in a row over time. But the researchers found out that these models have a secret: they’re better at understanding slow movements than fast ones! They also showed how to make these models better or worse at understanding fast movements by adjusting some settings. This can help them do a better job of learning things that are related over long periods. |
Keywords
» Artificial intelligence » Deep learning
