liveBook · Manning

Personalization

As explained in Applied Reinforcement Learning

Personalization is the adaptation of decisions, recommendations, or treatments to individual users or contexts.

Personalization adapts one-step decisions to context because the option that performs best on average may not be best for an individual. In contextual bandits, it works alongside continuous exploration and adaptation to changing customer behavior and markets. section 5.6

Practical design principles include:

  • Use context rather than relying only on population averages. section 5.6
  • Keep exploring so new options receive opportunities and unexpected changes do not go unnoticed. section 5.6
  • Plan for drift because customer behavior, competition, and markets change over time. section 5.6
  • Address cold starts with optimistic priors or warm starts that borrow knowledge from similar users or options. section 5.6

How Applied Reinforcement Learning covers it

How deep it goes
Connects personalization to context, exploration, cold starts, and drift but stops short of implementation detail.
Examples
Uses new ad creatives and newly acquired customer types to illustrate cold-start personalization challenges.
What you can do after
Apply it via contextual bandits by personalizing decisions, preserving exploration, and planning for drift.
How it connects
Connects personalization to contextual bandits, exploration–exploitation, cold starts, warm starts, and concept drift.

Also explained in

Compare how 11 books explain personalization →

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Applied Reinforcement Learning. The section links point to the parts of the book it draws on.