Teaching Curiosity-Driven AI to Play Games
Subscribe:  iCal  |  Google Calendar
San Antonio US   18, Jan — 21, Jan
London GB   21, Jan — 23, Jan
Taipei TW   24, Jan — 29, Jan
Zürich CH   31, Jan — 4, Feb
Leamington Spa GB   31, Jan — 3, Feb
Latest comments
by Protopop
5 hours ago

I love World Creator, especially the vegetation distribution pipeline. You can create some very realistic fields with it. Im going to check out impostors too - ive seen it a few times and wondered what it's about.

by Steve W
7 hours ago

Stupid Rental. My Zbrush that I haven't paid for an update in years and is perpetual Trumps that on all counts.

by Nameku
9 hours ago


Teaching Curiosity-Driven AI to Play Games
25 June, 2018

Take a look at another work that explains a way we can use AI to play games. The thing is that this one focuses on curiosity. How can a machine be curious? In this case, curiosity is defined by the AI’s ability to predict the results of its own actions. And this is big because the AI has the tools to acquire skills that don’t seem necessary now but might be in the future. 

First, let’s start by check it out a video on the paper by Two Minute Papers:


In many real-world scenarios, rewards extrinsic to the agent are extremely sparse or absent altogether. In such cases, curiosity can serve as an intrinsic reward signal to enable the agent to explore its environment and learn skills that might be useful later in its life. We formulate curiosity as the error in an agent’s ability to predict the consequence of its own actions in a visual feature space learned by a self-supervised inverse dynamics model. Our formulation scales to high-dimensional continuous state spaces, like images, bypasses the difficulties of directly predicting pixels, and, critically, ignores the aspects of the environment that cannot affect the agent. The proposed approach is evaluated in two environments: VizDoomand Super Mario Bros. Three broad settings are investigated: 1) sparse extrinsic reward, where curiosity allows for far fewer interactions with the environment to reach the goal; 2) exploration with no extrinsic reward, where curiosity pushes the agent to explore more efficiently; and 3) generalization to unseen scenarios (e.g. new levels of the same game) where the knowledge gained from earlier experience helps the agent explore new places much faster than starting from scratch. 

The paper “Curiosity-driven Exploration by Self-supervised Prediction” and its source code can be found here.

Leave a Reply

Be the First to Comment!