DEMO PAPER: Deep Reinforcement Learning for Real-Time Fear Induction in an SCP-087-Inspired Horror Game

This demo paper presents a novel approach to inducing fear in a horror game inspired by the SCP-087 creepypasta. The game features an infinite staircase that players must descend, with horror events triggered by real-time valence arousal values, as measured by facial expression detection. Deep reinf...

Full description

Saved in:
Bibliographic Details
Main Authors: Tu, Yuan, Mozgovoy, Maxim
Format: Conference Proceeding
Language:English
Subjects:
Deep learning
Deep reinforcement learning
facial expression detection
Games
horror game
Neural networks
Prototypes
real-time fear induction
Real-time systems
Reinforcement learning
SCP-087
Time factors
valence arousal values
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:This demo paper presents a novel approach to inducing fear in a horror game inspired by the SCP-087 creepypasta. The game features an infinite staircase that players must descend, with horror events triggered by real-time valence arousal values, as measured by facial expression detection. Deep reinforcement learning is used to train an AI agent to trigger these events and maintain players' valence arousal values at a certain level. The approach uses recent valence arousal values and game environment features as inputs to a neural network that outputs actions for the AI agent. As time constraints limited our ability to conduct thorough experiments, we present a prototype of our approach in this paper. While the results are preliminary, we believe that our prototype demonstrates the potential of our approach to induce fear in players using real-time facial expression detection and deep reinforcement learning. Future work will involve more extensive testing and refinement of our approach.
ISSN:2325-4289
DOI:10.1109/CoG57401.2023.10333157