Cerrisian Blog

I’m getting an itch to work on Antipathy.

This time, I’m using a slightly different strategy:

• PyMunk for collision detection
• cocos2d or grease for the game framework (pyglet for graphics)
• Twisted (of course) for networking/eventing
• Humane interface (with zooming) as much as possible!

Besides that I’m trying to simplify the game design and development. I just want some graphics, then networking and then go from there.

Until then, though, I’ll be running through tutorials over and over and over again.

A reinforcement learning based chat bot can be implemented by using the context free grammars created by the sequitur algorithm.

As phrases are learned by the chat bot, the sequitar algorithm creates grammar sub trees based on the existing corpus. These sub trees are cataloged and can be used as a state space. Simplistically, a state consists of a vector equal in length to the number of sub trees, with a bit flipped for whether that specific subtree was seen in the input.

Based on the state, various actions can be taken. New phrases could potentially be generated using the same grammar model, or other actions could be taken.

Reinforcement “simulations” or dreams could be executed between each user input.

Primary difficulties: Keep states consistent as new additions to the state space are added, Generating new phrases based on the captured grammar.