Understanding HTM

The model Today, I covered half of the Htm Paper by Cui et al 2016. A very interesting paper indeed.

So the problem the paper is trying to address is the following :

If we wanted to do sequence learning for real-time streaming data how would we approach this problem ?

It turns out the key is to open your mind :) ...
In order to get closer to the way the brain works Cui et al first defined what we deem more desirable in a sequence learning algorithm for such real time data which goes like the following :

  1. Continuous Learning : we need the algorithm to be able to learn continuously with the incoming data . now in ML, this is done in two main ways. one is to go with batch learning algorithm where we have a predefined buffer dataset from past data. This method is very resource intensive. The other is online learning which basically means the model learns as it receives incoming data.

to be continued ...

References

Cui Y, Ahmad S, Hawkins J. Continuous Online Sequence Learning with an Unsupervised Neural Network Model. Neural Comput. 2016 Nov;28(11):2474-2504. doi: 10.1162/NECO_a_00893. Epub 2016 Sep 14. PMID: 27626963. https://pubmed.ncbi.nlm.nih.gov/27626963/