The Sea Creature Expert Robot!
It's time to teach the robot about all sea creatures 🐸🐡🦑
Our robot learns patterns! 🧐
A pattern means a «repeating shape» in the pictures of fish (the data).
AI identifies these repetitions so it can make predictions or decisions.
What do you think the robot considers as patterns to recognize sea creatures?
Well done! 🎉😆 All the options are correct. The robot sees all of these and learns them. These things, like color, texture, body shape, and special features like having a tail or fins, are found in most fish; so they count as patterns for our robot!
So far we've learned that our robot is really good at learning repeating patterns. When we show it different fish, it pays attention to the patterns that keep repeating. For example, all fish have eyes, all fish have tails, fish are roughly similar in size and color, and so on...
But does it matter how many fish our robot was trained with? The more fish it sees, does it recognize the patterns better?
Is Hoosha's robot better or Juia's robot?
Hoosha and Juia are two friends who want to play the «Clean the Ocean» game.
Juia trains the robot with only 5 pictures of trash and sea creatures. But Hoosha does it with 50 pictures. Which robot can clean the ocean better, Hoosha's or Juia's? (Juia and Hoosha are both good teachers!)
Correct answer: Hoosha's robot. Because it was trained more.
The more we train the robot, the better it learns, and as a result it performs better. That means it can separate sea creatures from trash really well.