Hopfield/Brody mouse (mus silicium)

Time-warp-invariant computation with action potentials

In December 2000, the Inference Group won both parts of Hopfield and Brody's `mouse brain' competition.

For competition A, the challenge was to explain how a simulated `mouse brain' made up of about a thousand neurons performed speech recognition; competition B required entrants to construct their own simulated brain, capable of speech recognition on a ten-word vocabulary, like Hopfield and Brody's.

• Press release about our winning entries
• Here | Hopfield & Brody's website about the mouse and the competition
• Our submission to competition A, which gives our deductions about how mus silicium works
• An MPEG movie (.asf) of David MacKay describing our theory with the aid of two slices of pizza
• Seb Wills's winning entry to competition B: description and results
• Background facts: How many neurons does it take to do speech recognition?

David MacKay & Seb Wills
14th December 2000

Contact information:

David MacKay: +44-1223-339852, or via http://www.inference.org.uk/mackay/
Seb Wills: +44-1223 337344, or via http://www.inference.org.uk/saw27/