Decision Loss

Decision loss focuses on aligning machine learning models with downstream optimization tasks, aiming to improve the quality of decisions made using model predictions rather than solely maximizing prediction accuracy. Current research emphasizes developing methods that integrate machine learning and optimization seamlessly, often employing techniques like mixture-of-experts models, differentiable optimization, and surrogate loss functions to address challenges such as computational cost and non-differentiable optimization problems. This approach holds significant promise for improving decision-making in diverse fields, including marketing, resource allocation, and reinforcement learning, by ensuring that models are trained to directly benefit the ultimate decision-making process.

Papers

July 18, 2024

Decision Focused Causal Learning for Direct Counterfactual Marketing Optimization
Hao Zhou, Rongxiao Huang, Shaoming Li, Guibin Jiang, Jiaqi Zheng, Bing Cheng, Wei Lin
Decision Focused Learning Decision Loss Marketing Optimization Step Optimization

July 16, 2024

Scaling Diffusion Transformers to 16 Billion Parameters
Zhengcong Fei, Mingyuan Fan, Changqian Yu, Debang Li, Junshi Huang
Diffusion Model Many Parameter Diffusion Transformer Dense Network Conditional Image Generation Decision Loss

May 23, 2024

Decision-Focused Forecasting: Decision Losses for Multistage Optimisation
Egon Peršak, Miguel F. Anjos
State of the Art Forecasting Energy Storage Differentiable Optimization Multistage Stochastic Decision Loss

December 6, 2023

Enhancing Molecular Property Prediction via Mixture of Collaborative Experts
Xu Yao, Shuang Liang, Songqiao Han, Hailiang Huang
Mixture Component Molecular Graph Molecular Property Prediction Multiple Expert Multidimensional Projection Decision Loss

July 18, 2023

Landscape Surrogate: Learning Decision Losses for Mathematical Optimization Under Partial Information
Arman Zharmagambetov, Brandon Amos, Aaron Ferber, Taoan Huang, Bistra Dilkina, Yuandong Tian
Optimization Purpose General Purpose Landscape Analysis Incomplete Information Numerical Optimization Learning Based Optimization Decision Loss

June 30, 2023

$\lambda$-models: Effective Decision-Aware Reinforcement Learning with Latent Models
Claas A Voelcker, Arash Ahmadian, Romina Abachi, Igor Gilitschenski, Amir-massoud Farahmand
Model Based Reinforcement Learning Latent Variable Model Lambda Calculus Decision Loss Decision Aware

November 15, 2022

Decision-Aware Learning for Optimizing Health Supply Chains
Tsai-Hsuan Chung, Vahid Rostami, Hamsa Bastani, Osbert Bastani
Optimization Purpose Supply Chain Classification Error Downstream Decision Making Decision Loss Decision Aware

October 30, 2022

Learning to Defer to Multiple Experts: Consistent Surrogate Losses, Confidence Calibration, and Conformal Ensembles
Rajeev Verma, Daniel Barrejón, Eric Nalisnick
LeArning Abstract Confidence Calibration Multiple Expert Surrogate Loss Decision Loss One v the Rest Loss Consistent Surrogate Conformal Meta Learner

August 7, 2022

Optimal Tracking in Prediction with Expert Advice
Hakan Gokcesu, Suleyman S. Kozat
Human Prediction Regret Guarantee Min Max Decision Loss Expert Advice Problem

March 30, 2022

Decision-Focused Learning without Differentiable Optimization: Learning Locally Optimized Decision Losses
Sanket Shah, Kai Wang, Bryan Wilder, Andrew Perrault, Milind Tambe
LeArning Abstract Differentiable Optimization Decision Focused Learning Downstream Decision Making Decision Loss Operator Surrogate

Decision Loss

Papers

Decision Focused Causal Learning for Direct Counterfactual Marketing Optimization

Scaling Diffusion Transformers to 16 Billion Parameters

Decision-Focused Forecasting: Decision Losses for Multistage Optimisation

Enhancing Molecular Property Prediction via Mixture of Collaborative Experts

Landscape Surrogate: Learning Decision Losses for Mathematical Optimization Under Partial Information

$\lambda$-models: Effective Decision-Aware Reinforcement Learning with Latent Models

Decision-Aware Learning for Optimizing Health Supply Chains

Learning to Defer to Multiple Experts: Consistent Surrogate Losses, Confidence Calibration, and Conformal Ensembles

Optimal Tracking in Prediction with Expert Advice

Decision-Focused Learning without Differentiable Optimization: Learning Locally Optimized Decision Losses