Decoding of brain information
Abstract
Since the early 2000s, researchers have been actively studying brain information decoding, which reads perceptual and cognitive contents from brain activities. Recent studies reconstructed the images seen by subjects from their brain activities and read out the content of dreams (both studies were done by the Kamitani Group of ATR).
These were made possible thanks to the development of machine learning, which automatically learns the relationship between brain activity patterns (generally spatiotemporal patterns) and perceptual states from data.
We proposed a machine learning method suitable for the nature of brain activity data (low signal-to-noise ratio, huge dimensions, and limited samples).
We released open source software implementing our proposed method, and this software has been used in various neuroscience studies.
Download:
The sparse estimation library