Teach AI to Turn Lights On and Off in 15 Lines of Code: A Smart Energy-Saving Butler with Neural Networks Linux爱好者

15 Lines of Code to Teach AI to Turn Lights On and Off: A Smart Energy-Saving Butler with Neural Networks

No rules, no if-else, just “trial and error + reward” — let AI learn to intelligently control lighting. This is the magic of Reinforcement Learning!

Is your home light still manually operated? Or using a simple “someone’s there, turn on; no one’s there, turn off” logic?

Reality is more complex:

Daytime, someone’s present, but plenty of sunlight → no need for lights, save power!
Late night, someone’s present, room is dark → must turn on lights, prioritize experience!
No one around → lights on is wasteful!

Can the system learn to make optimal decisions in various situations?

The answer is yes! Using reinforcement learning + neural networks.

It needs to understand three pieces of information (State):

Then decide on an Action:

Its “teacher” uses only one number — Reward — to tell it: “You did well, +1 point!” or “Wasted electricity, -0.3 points!”

Over time, the AI learns a strategy that’s both energy-efficient and considerate.

Traditional approach: write lots of if-else
RL approach: Let the neural network learn the “thresholds” and “logic” itself!

Input state → Neural network → Output each action’s “expected value” (Q-value) → Choose the highest-value action

This is the essence of Deep Q-Network (DQN).

No labels, no supervision, only reward signals for self-evolution!

Teach AI to Turn Lights On and Off in 15 Lines of Code: A Smart Energy-Saving Butler with Neural Networks