Tutorial on Amortized Optimization

DOI10.1561/2200000102zbMATH Open1525.68107arXiv2202.00665MaRDI QIDQ6139544FDOQ6139544

Brandon D. Amos

Publication date: 19 December 2023

Published in: Foundations and Trends® in Machine Learning (Search for Journal in Brave)

Abstract: Optimization is a ubiquitous modeling tool and is often deployed in settings which repeatedly solve similar instances of the same problem. Amortized optimization methods use learning to predict the solutions to problems in these settings, exploiting the shared structure between similar problem instances. These methods have been crucial in variational inference and reinforcement learning and are capable of solving optimization problems many orders of magnitudes times faster than traditional optimization methods that do not use amortization. This tutorial presents an introduction to the amortized optimization foundations behind these advancements and overviews their applications in variational inference, sparse coding, gradient-based meta-learning, control, reinforcement learning, convex optimization, optimal transport, and deep equilibrium networks. The source code for this tutorial is available at https://github.com/facebookresearch/amortized-optimization-tutorial.

Full work available at URL: https://arxiv.org/abs/2202.00665

zbMATH Keywords

optimizationdeep learningreinforcement learning

Mathematics Subject Classification ID

Numerical optimization and variational techniques (65K10) Learning and adaptive systems in artificial intelligence (68T05) Research exposition (monographs, survey articles) pertaining to computer science (68-02) Artificial neural networks and deep learning (68T07) Research exposition (monographs, survey articles) pertaining to numerical analysis (65-02)

Cites Work

• Title not available (Why is that?)
• Title not available (Why is that?)
• Title not available (Why is that?)
• Title not available (Why is that?)
• Title not available (Why is that?)
• Title not available (Why is that?)
• Title not available (Why is that?)
• Title not available (Why is that?)
• Title not available (Why is that?)
• Title not available (Why is that?)
• Title not available (Why is that?)
• Title not available (Why is that?)
• A Fast Iterative Shrinkage-Thresholding Algorithm for Linear Inverse Problems
• CVXPY: a Python-embedded modeling language for convex optimization
• Julia: A Fresh Approach to Numerical Computing
• JuMP: A Modeling Language for Mathematical Optimization
• OSQP: An Operator Splitting Solver for Quadratic Programs
• Graphical Models, Exponential Families, and Variational Inference
• Elements of Information Theory
• Distributed Optimization and Statistical Learning via the Alternating Direction Method of Multipliers
• Conic optimization via operator splitting and homogeneous self-dual embedding
• Intrinsic statistics on Riemannian manifolds: Basic tools for geometric measurements
• Optimal transport for applied mathematicians. Calculus of variations, PDEs, and modeling
• Machine learning. A probabilistic perspective
• Optimally sparse representation in general (nonorthogonal) dictionaries via ℓ ¹ minimization
• Anderson Acceleration for Fixed-Point Iterations
• Iterative Procedures for Nonlinear Integral Equations
• \({\mathcal Q}\)-learning
• Implicit Functions and Solution Mappings
• An iterative thresholding algorithm for linear inverse problems with a sparsity constraint
• Envelope Theorems for Arbitrary Choice Sets
• A tutorial on the cross-entropy method
• A Class of Methods for Solving Nonlinear Simultaneous Equations
• The Theory of Max-Min, with Applications
• An introduction to variational methods for graphical models
• Atomic decomposition by basis pursuit
• OnActor-Critic Algorithms
• Sensitivity analysis of generalized equations
• Convex optimization: algorithms and complexity
• Modern homotopy methods in optimization
• Data-Driven Nonsmooth Optimization
• Sensitivity and stability analysis for nonlinear programming
• The principle of maximum entropy
• Machine learning for combinatorial optimization: a methodological tour d'horizon
• A Flexible and Efficient Algorithmic Framework for Constrained Matrix and Tensor Factorization
• Lectures on convex optimization
• Mathematics for Machine Learning
• On learning and branching: a survey
• End-to-end training of deep visuomotor policies
• A brief introduction to manifold optimization
• An Introduction to Variational Autoencoders
• The voice of optimization
• Large scale model predictive control with neural networks and primal active sets
• Online Mixed-Integer Optimization in Milliseconds
• Solving and learning nonlinear PDEs with Gaussian processes
• Accelerated optimization on Riemannian manifolds via discrete constrained variational integrators
• Solution refinement at regular points of conic problems
• Learning to predict the cosmological structure formation
• Globally Convergent Type-I Anderson Acceleration for Nonsmooth Fixed-Point Iterations

This page was built for publication: Tutorial on Amortized Optimization

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6139544)