DeepAI AI Chat
Log In Sign Up

Optimization with Gradient-Boosted Trees and Risk Control

by   Miten Mistry, et al.

Decision trees effectively represent the sparse, high dimensional and noisy nature of chemical data from experiments. Having learned a function from this data, we may want to thereafter optimize the function, e.g., picking the best chemical process catalyst. In this way, we may repurpose legacy predictive models. This work studies a large-scale, industrially-relevant mixed-integer quadratic optimization problem involving: (i) gradient-boosted pre-trained regression trees modeling catalyst behavior, (ii) penalty functions mitigating risk, and (iii) penalties enforcing composition constraints. We develop heuristic methods and an exact, branch-and-bound algorithm leveraging structural properties of gradient-boosted trees and penalty functions. We numerically test our methods on an industrial instance.


page 1

page 2

page 3

page 4


Improved Penalty Algorithm for Mixed Integer PDE Constrained Optimization (MIPDECO) Problems

Optimal control problems including partial differential equation (PDE) a...

Quant-BnB: A Scalable Branch-and-Bound Method for Optimal Decision Trees with Continuous Features

Decision trees are one of the most useful and popular methods in the mac...

Network Optimization via Smooth Exact Penalty Functions Enabled by Distributed Gradient Computation

This paper proposes a distributed algorithm for a network of agents to s...

Mixed-Integer Optimization with Constraint Learning

We establish a broad methodological foundation for mixed-integer optimiz...

Mixed integer linear optimization formulations for learning optimal binary classification trees

Decision trees are powerful tools for classification and regression that...

An interpretable machine learning approach for ferroalloys consumptions

This paper is devoted to a practical method for ferroalloys consumption ...

Optimal Decision Trees for the Algorithm Selection Problem: Integer Programming Based Approaches

Even though it is well known that for most relevant computational proble...