Many kinds of machine learning challenges - like those involving static computer vision tasks - deal with datasets whose members are un-ordered or static in nature. That is, as the data is generated individual points just sort of 'pop' into existence at random.
However many types of data arise sequentially in nature, in an ordered fashion, often when data is generated in time. This sort of data is collectively referred to as time series.
Financial data is a common example: the price of stocks, commodities, etc., are all ‘time-stamped’ with a date (the price of stock X was Y at date Z).
Speech recognition - baked into most smartphones and tablets these days - gives software a verbal user interface. The data used to train the AI for this technology are snippets of spoken sentences, with the words (in text) marked on each snippet (as illustrated in the image above)..
Data from electric machinery measuring voltages, motion sensors, data from servos, and camera feeds are all naturally time series.
Patients' medical history recording doctors visits, trips to the hospital, and other medical procedures are naturally time series as well.
Does your business generate time series data like the examples listed above? Reach out and let's talk about how AI can put your time series data to work for your business.