Project Overview
Build an Agent based on Simulated Neurons
- The agent is based on simulated neurons. We're hoping to do
it with our existing fatiguing Leaky Integrate and Fire model.
This is driven from the Cell Assembly hypothesis.
- The agent is situated in a video game environment, and
interacts with a user in this same environment. It will
probably be an assistant.
- The agent views its surrounding based on the video game
images it sees. These are the same the user sees though
from a different perspective.
- The agent takes commands in Natural Language.
- The agent has its own goals and behavioural mechanisms.
The goals can be set by the NL commands from the user.
- This is an engineering problem. However, it has ramifications
for neurophysiology, psychology, and the full range of AI tasks.
- If it did things just like people did, we'd be entirely chuffed.
One way to subdivide the agent is:
- Vision
- Natural Language Processing
- Action
