Task-based End-to-end Model Learning in Stochastic Optimization

Abstract

With the increasing popularity of machine learning techniques, it has become common to see prediction algorithms operating within some larger process. However, the criteria by which we train these algorithms often differ from the ultimate criteria on which we evaluate them. This paper proposes an end-to-end approach for learning probabilistic machine learning models in a manner that directly captures the ultimate task-based objective for which they will be used, within the context of stochastic programming. We present three experimental evaluations of the proposed approach: a classical inventory stock problem, a real-world electrical grid scheduling task, and a real-world energy storage arbitrage task. We show that the proposed approach can outperform both traditional modeling and purely black-box policy optimization approaches in these applications.

arXiv commentIn NIPS 2017. Code available at https://github.com/locuslab/e2e-model-learning

Cited by in corpus (38)

• DC3: A learning method for optimization with hard constraints
• End to end learning and optimization on graphs
• ENTMOOT: A Framework for Optimization over Ensemble Tree Models
• SATNet: Bridging deep learning and logical reasoning using a differentiable satisfiability solver
• Decision Trees for Decision-Making under the Predict-then-Optimize Framework
• How to Train Your Differentiable Filter
• Interior Point Solving for LP-based prediction+optimisation
• Learning to Solve Network Flow Problems via Neural Decoding
• Improving Deep Learning Models via Constraint-Based Domain Knowledge: a Brief Survey
• Objective Mismatch in Model-based Reinforcement Learning
• Particle Filter Networks with Application to Visual Localization
• Gradient-Aware Model-based Policy Search
• Pontryagin Differentiable Programming: An End-to-End Learning and Control Framework
• Melding the Data-Decisions Pipeline: Decision-Focused Learning for Combinatorial Optimization
• Learning Linear Programs from Optimal Decisions
• Distributionally robust stochastic programs with side information based on trimmings -- Extended version
• The Differentiable Cross-Entropy Method
• Implicit MLE: Backpropagating Through Discrete Exponential Family Distributions
• Fast Rates for Contextual Linear Optimization
• Data Sharing and Compression for Cooperative Networked Control
• Automatically Learning Compact Quality-aware Surrogates for Optimization Problems
• Risk Bounds and Calibration for a Smart Predict-then-Optimize Method
• Integrating Algorithmic Planning and Deep Learning for Partially Observable Navigation
• Adversarially Robust Learning for Security-Constrained Optimal Power Flow
• Heteroscedasticity-aware residuals-based contextual stochastic optimization
• Differentiable Implicit Layers
• Understanding Deep Architectures with Reasoning Layer
• Balanced Order Batching with Task-Oriented Graph Clustering
• Deep Inverse Optimization
• A Surrogate Objective Framework for Prediction+Optimization with Soft Constraints
• Task-Based Learning via Task-Oriented Prediction Network with Applications in Finance
• Learning MDPs from Features: Predict-Then-Optimize for Sequential Decision Problems by Reinforcement Learning
• Task-aware Privacy Preservation for Multi-dimensional Data
• Accelerating Goal-Directed Reinforcement Learning by Model Characterization
• LfEdNet: A Task-based Day-ahead Load Forecasting Model for Stochastic Economic Dispatch
• Integrated Optimization of Predictive and Prescriptive Tasks
• Divide and Learn: A Divide and Conquer Approach for Predict+Optimize
• Differentiable Greedy Submodular Maximization: Guarantees, Gradient Estimators, and Applications