What is signal detection theory?

Signal detection theory is the measure of distinction between signals that carry information and signals that are just noise. Measuring the difference is important to make sure decisions are based on actual information, and not random noise or minimizing the amount of uncertainty involved in decision making. In the healthcare field specifically, this is crucial to ensure hospital staff can tell the difference between specific alarms and act on that information.

I conducted my own signal detection experiment on myself. You can read about it here if you're interested.

Tesla's computer vision

It's interesting to think about signal detection theory in conjunction with automation. Ideally, robots shouldn't make decisions that contain the same level of uncertainty as humans because they should have better signal processing than humans. However, as more and more systems become automated, we can see that robots sometimes need more training that people. For example, Tesla trains its computer vision system to differentiate between trees, road, cars, humans, bikes. The computer is still making decisions based on the amount of certainty behind each of these identifications. It would be interesting to better understand what kind of analysis Tesla does behind the scenes.

Tesla's computer vision identifying objects in Paris.