Black-Box Optimization Challenge: “Hidden Oracle”

In this project, you will interact with a hidden black-box system through a provided executable. This can be found here:

oracle.exe

The system takes five input parameters:

x1, x2, x3, x4, x5

Each parameter is in the range:

0 ≤ xi ≤ 1

The system returns a single scalar output y representing the performance of that input design.

You cannot see the function behind the oracle — it is intentionally hidden and noisy. This mimics real-world situations where:

• the underlying system is unknown,
• measurements are noisy,
• and evaluations are expensive or limited.

Objective

Your goal is to find the input combination that maximizes y.

Constraints:

• The maximum number of evaluations to find the optimal solution is 60
• Each evaluation consumes one or more rows from your input CSV.

You will need to design a strategy that efficiently explores the input space, learns from past observations, and identifies promising designs.