Summary of Graph Agnostic Causal Bayesian Optimisation, by Sumantrak Mukherjee et al.
Graph Agnostic Causal Bayesian Optimisation
by Sumantrak Mukherjee, Mengyan Zhang, Seth Flaxman, Sebastian Josef Vollmer
First submitted to arxiv on: 5 Nov 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Machine Learning (stat.ML)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The paper investigates Causal Bayesian Optimization (CBO), a problem that involves optimizing a target variable on an unknown causal graph, with various interventions possible. The authors formalize this issue as CBO under cumulative regret objectives for two settings: structural causal models with hard interventions and function networks with soft interventions. They propose Graph Agnostic Causal Bayesian Optimization (GACBO), which actively discovers the causal structure contributing to optimal rewards. GACBO balances exploiting the best actions with exploring the causal structures and functions. This work is the first to study cumulative regret objectives in scenarios where the graph is unknown or partially known, outperforming baselines in simulated experiments and real-world applications. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Causal Bayesian Optimization (CBO) is a way to find the best solution for a problem that depends on many factors. The problem involves a “target variable” that we want to optimize, but we don’t know how it’s affected by different variables. This paper looks at two ways to approach this issue: when we can intervene in certain ways (hard interventions) and when we can only nudge things slightly (soft interventions). They come up with an algorithm called Graph Agnostic Causal Bayesian Optimization (GACBO), which tries to find the best solution by balancing what works well now with exploring new possibilities. This is a new way of doing things that can help us solve problems better in situations where we’re not sure how everything affects each other. |
Keywords
* Artificial intelligence * Optimization
