issues
search
breznak
/
neural.benchmark
Comparisons of HTM to other ML algorithms on well known datasets and synthetic anomaly benchmarks
GNU General Public License v2.0
8
stars
1
forks
source link
Improved contectual memory (TM)
#46
Open
breznak
opened
8 years ago
breznak
commented
8 years ago
[ ] TM grows number of context-dedicated neurons the longer we follow a known sequence, #45 #41
[ ] SP/TM reflects "importance"/frequency of concepts in predictions:
train: A,B A,B A,C predict: A,? ....?=B(66%)&C(33%)
[ ] multiple scales/zooms at the same time; locality vs. globality, #42
[ ] TM outputs SDR (with all the nice properties-robustness, distributed,sparse)
currently inhibition is not ran on TP level!
nor is boosting, concepts known from SP
is this fixed is TM?