Summary of Scaling Up Ridge Regression For Brain Encoding in a Massive Individual Fmri Dataset, by Sana Ahmadi and Pierre Bellec and Tristan Glatard
Scaling up ridge regression for brain encoding in a massive individual fMRI dataset
by Sana Ahmadi, Pierre Bellec, Tristan Glatard
First submitted to arxiv on: 28 Mar 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI); Neurons and Cognition (q-bio.NC); Quantitative Methods (q-bio.QM)
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 This paper investigates the efficiency of parallelization techniques to speed up training time for brain encoding with ridge regression on large-scale deep functional magnetic resonance imaging (fMRI) datasets. The researchers focus on the CNeuroMod Friends dataset, a prominent resource in this field, and explore different approaches to reduce training time. They compare multi-threading using Intel Math Kernel Library (MKL) to OpenBLAS library, finding that MKL outperforms OpenBLAS by 1.9 times when running 32 threads on a single machine. The study also evaluates the Dask multi-CPU implementation of ridge regression and proposes a new “batch” version motivated by time complexity analysis. The results show that while MultiOutput parallelization is impractical, the Batch-MultiOutput regression scales well across compute nodes and threads, providing significant speed-ups. This work highlights the potential of batch parallelization using Dask for brain encoding with ridge regression on high-performance computing systems. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper explores ways to make a computer program faster when it’s trying to understand human brain activity from movies or other images. The researchers are looking at how to speed up this process by using different techniques, like splitting the task into smaller parts and doing them all at once. They tested these techniques on a large dataset of brain activity information and found that one approach was much faster than others. This work could help scientists study brain activity more efficiently and make new discoveries. |
Keywords
* Artificial intelligence * Regression
