for rob

[mobeets]

For Rob:

less motivation, but still make it clear more math keep same structure: basic hyps, results, new hyps, results learning hyps by around 21 mins

make sure to walk through chain of logic:

neurons -> factors -> null space -> rotation

when presenting, maybe foreshadow the cloud as being "now here's where we're GOING to get," but don't discuss how. that way when you show minimal energy hyps, the audience won't care as much

[mobeets]

stats:

• [ ] is there significant tuning?
• [ ] mean/covariance errors lower for cloud than hab (sign test, or wilcoxon signed-rank test?)