OPERNEX LEARNING

Learn Agentic AI

A structured, practical learning path from language model foundations to intelligent agentic systems.

6 Learning parts
453 Subtopics
415 Published lessons
Foundations / Introduction to Reinforcement Learning / Core Concepts and Definitions
Lesson 105 of 415 6 min read
3.2.2

Return and Cumulative Discounted Reward

Understand how rewards are accumulated over time and how the discount factor affects future rewards.

What is it?

Return is the cumulative reward collected over time. Discounted return is expressed as Gₜ = Σₖ₌₀∞ γᵏ rₜ₊ₖ. The discount factor controls the importance of future rewards.

Key points

1. Return represents cumulative reward. 2. Rewards are collected at each step. 3. Future rewards are discounted using γ. 4. Generally, 0 ≤ γ ≤ 1. 5. Smaller γ emphasizes immediate rewards. 6. Larger γ gives greater importance to future rewards.

Example: Calculating discounted return

Suppose an agent receives: r₀ = 10 r₁ = 5 r₂ = 2 Let γ = 0.9. G₀ = 10 + (0.9 × 5) + (0.9² × 2) G₀ = 10 + 4.5 + 1.62 G₀ = 16.12

Real-world application

Discounted return defines the objective for agents balancing immediate and long-term rewards.
Key takeaway

Return measures accumulated reward; the discount factor controls the importance of future rewards.