I have learned that the HTM system is suitable for processing the streaming data in real time, I want to know
(1) if the system should to run for a very long time (for the industry purpose), if the storage of patterns learned by Spacial Pooling is full, what will happen? will it happen in the future?
(2) How to deal with that?
(3) Is there any mechanism to prevent the situation?
(4) Do I need to worry about that?
Thanks a lot!
How long can the HTM keeps learning and processing data for? Might overflow after a long enough time running?
Hey there =)
I don’t think that you have to worry. In the Paper about SDRs are some interesting properties of SDRs mentioned. One is the capacity (formula (1))… so how many unique SDR encodings are possible with given n (total length of SDR) and w (number of on bits).
In fact: if you use a common setup for your SP with 2048 total Columns and 40 active columns in every time step, you can encode 2.37×10^84 different values.
This number is suuuuper large
So I would say: no worries!