In the email Wyze sent out today… in the “Wyze AI Model Updates” section, they show how challenging it is for the AI to tell the difference between faces of a Chihuahua and that of blueberry muffins.
Is the reason why Wyze wants the AI to know the difference because they want the AI to NOT trigger an alert if it sees your Chihuahua walking around the house, but the ai SHOULD trigger an alert if a blueberry muffin is walking around the house?
I is almost comparable to the human brain and a phenomenon called pareidolia.
for everyone learning a new word
pareidolia
: the tendency to perceive a specific, often meaningful image in a random or ambiguous visual pattern
The scientific explanation for some people is pareidolia, or the human ability to see shapes or make pictures out of randomness. Think of the Rorschach inkblot test.
think of your parents telling you about “the man on the moon” or seeing a face in the pattern of tree bark. the same can happen with machine learning where attributes can be falsely given meaning to the wrong things with adverse consequences to the learning model. this is one reason why your training base data needs to be so wide and encompassing but also clear about what exactly it is seeing.
and yes…it absolutely should be triggering an event for that!!! I WANT TO SEE IT!
that can be the next filter for the AI…“walking muffin”
Yep, the training of the AI is more about building a library or database that includes almost but not quite the Chihuahua. Its as much that the AI needs to know what is not the dog than is the dog.