This is a kaggle winner who used simplistic neurons in a convoluted manner which resembles some aspects of L1-L6 neocortical structures and their interconnectivity.
Has anyone ever tried figuring out how particle decay occurs (with patterns which are very difficult to detect from background noise) using HTM, or specifically NuPIC ?
What encoders could be created for the collision patterns as input streams from ATLAS ?
If you were at CERN we could have a chat. What were the training data in this competition? Was temporal memory even applicable? (causal series)
I have no association with CERN or any other academic institution. Asking these questions is my alternative to looking through a telescope from my bedroom window at stars and saying “ohhhh - pretty”. Proto-Hobbiest.
The training data had many events, each with 30 recorded features, and a weight (for how “reliably” that information was captured) ? I think all the events occur in a single … time step. In this context I don’t understand the word “time” entirely, so here is the original competition:
training.csv - Training set of 250000 events, with an ID column, 30 feature columns, a weight column and a label column.
test.csv - Test set of 550000 events with an ID column and 30 feature columns.
It seems CERN do, or have used specialized hardware based associative memory for event detection at high speed.