I was wondering how an AI algorithm for real-time sequential (e.g. a flow sensor or heat-beat sensor) analysis is working.
With a classic algorithm I can analyze each individual new data point, for example with a peak finding algorithm. I can compare it to a threshold and can raise an alarm or measure the time the signal stays above the threshold (number of data points).
How does it work, when I want to implement an AI for analyzing the same signal, because I want to exploit the shape of the signal variation as well (like it's done in a heart beat anomaly detection)? Do I still feed every single new data point? Or do I use a sliding window?
Where can I read up on that topic?
