Deep neural networks and mixed integer linear optimization

DOI10.1007/s10601-018-9285-6zbMath1402.90096arXiv1712.06174OpenAlexW2802557767MaRDI QIDQ1617390

Jason Jo, Matteo Fischetti

Publication date: 8 November 2018

Published in: Constraints (Search for Journal in Brave)

Full work available at URL: https://arxiv.org/abs/1712.06174

zbMATH Keywords

computational experimentsmixed-integer programmingmathematical optimizationdeep learningdeep neural networks

Mathematics Subject Classification ID

Mixed integer programming (90C11) Learning and adaptive systems in artificial intelligence (68T05)

Related Items

Exploiting verified neural networks via floating point numerical error ⋮ Modeling design and control problems involving neural network surrogates ⋮ The role of optimization in some recent advances in data-driven decision-making ⋮ Global optimization of objective functions represented by ReLU networks ⋮ Getting away with more network pruning: from sparsity to geometry and linear regions ⋮ OAMIP: optimizing ANN architectures using mixed-integer programming ⋮ Speeding up neural network robustness verification via algorithm configuration and an optimised mixed integer linear programming solver portfolio ⋮ On mathematical optimization for clustering categories in contingency tables ⋮ Data-driven robust optimization using deep neural networks ⋮ Fast BATLLNN: Fast Box Analysis of Two-Level Lattice Neural Networks ⋮ Optimization of sparsity-constrained neural networks as a mixed integer linear program ⋮ Deep Neural Networks Pruning via the Structured Perspective Regularization ⋮ Convex and concave envelopes of artificial neural network activation functions for deterministic global optimization ⋮ BERN-NN: Tight Bound Propagation For Neural Networks Using Bernstein Polynomial Interval Arithmetic ⋮ Data-driven optimization model customization ⋮ Optimization problems for machine learning: a survey ⋮ Classifier-based constraint acquisition ⋮ Advances in verification of ReLU neural networks ⋮ Strong mixed-integer programming formulations for trained neural networks ⋮ An outer-approximation guided optimization approach for constrained neural network inverse problems ⋮ A two-stage exact algorithm for optimization of neural network ensemble ⋮ Injecting domain knowledge in neural networks: a controlled experiment on a constrained problem ⋮ Between steps: intermediate relaxations between big-M and convex hull formulations

Uses Software

• CPLEX
• ImageNet
• AlexNet

Cites Work

• Proximity search for 0--1 mixed-integer convex programming
• Local branching
• Fast training of support vector machines with Gaussian kernel
• On handling indicator constraints in mixed integer programming
• An Evolutionary Algorithm for Polishing Mixed Integer Programming Solutions
• Unnamed Item

This page was built for publication: Deep neural networks and mixed integer linear optimization