This summer I will be participating in the Build-A-Bot Lab at the University of Denver. The goal of the Build-A-Bot Lab is to create an algorithm that is able to create robots that are perceived in a certain way by humans. For example, robots that work in the medical field should be perceived as caring and warm. What certain features create this perception in a robot? Robots that work in the army should be perceived as strict and tough. What certain features of the robot create this perception? This is what the Build-A-Bot Lab aims to answer.
How do we plan on answering this question?
Well, the Build-A-Bot Lab is creating a game where players are prompted to create a robot that shows an attribute (caring, vicious, hopeful, sad, hungry, etc.). The plan is through this game, we could get a lot of input and data that could be used to train a machine-learning algorithm that will then be able to determine what certain characteristics of a robot contribute to that certain attribute. We've teamed up with the Psychology department as well, and they have been trying to find how these perceptions are created in the human brain.
What is machine learning?
Machine learning is a form of artificial intelligence. It's an attempt to get a computer program to think for itself. Machine learning takes in input, and displays the output. With machine learning specifically, you "train" the algorithm by having a ton of data as input (training data) and telling the algorithm what the correct outputs are. Then you test the algorithm by showing it input that is similar to the training data but not identical to it, and then determine the success rate of the outputs of the algorithm. Once the algorithm is trained well enough that its success rate is high, it is a fully trained machine learning algorithm.
Example: Input (Data) -> Machine Learning Algorithm -> Output Image of Apple-> Machine Learning Algorithm -> Banana (INCORRECT) Image of Pear -> Machine Learning Algorithm -> Pear (CORRECT)
I've got a long way to go to understand the complexities of how the machine learning algorithm actually works behind the code. But for now, it's an honor to be a part of the team.
Comments