devslem / multiple-knapsack

Optimize multiple knapsack problem using reinforcement learning.

Profile 👋

University - Incheon National University
Blog - DevSlem Blog
E-mail - [email protected]
LinkedIn - Jinyeong Park

Optimize my daily routine using reinforcement learning (Q-learning):

# take a step
S = "busy" # state
A = "play" in { "play", "work" } # action
S' = "boom" # next state
R = -100 # penalty

# learn from the experience
Q[S,A] += alpha * (R + gamma * max(Q[S',:]) - Q[S,A])

# but, still I don't want to work...
A' = "play" # next action

Tech Stack

Field

• Reinforcement Learning (RL) 🔥
• Machine Learning

Project

• Back to the Dungeon: Unity 2D Platformer Shooting Game
  • Repository - only script files
  • Download
• AINE-DRL: Deep Reinforcement Learning Baseline Framework
  • Repository
  • Download
• Mol-AIR: Molecular Reinforcement Learning with Adaptive Intrinsic Reward for Goal-directed Molecular Generation
  • Repository
  • Paper
• Multiple Knapsack: Multiple Knapsack Optimization using Deep Reinforcement Learning with Transformer
  • Repository