M-HOF-Opt: Multi-Objective Hierarchical Output Feedback Optimization via Multiplier Induced Loss Landscape Scheduling
Not on the map. Why?
Abstract (arXiv)
A probabilistic graphical model is proposed, modeling the joint model parameter and multiplier evolution, with a hypervolume based likelihood, promoting multi-objective descent in structural risk minimization. We address multi-objective model parameter optimization via a surrogate single objective penalty loss with time-varying multipliers, equivalent to online scheduling of loss landscape. The multi-objective descent goal is dispatched hierarchically into a series of constraint optimization sub-problems with shrinking bounds according to Pareto dominance. The bound serves as setpoint for the low-level multiplier controller to schedule loss landscapes via output feedback of each loss term. Our method forms closed loop of model parameter dynamic, circumvents excessive memory requirements and extra computational burden of existing multi-objective deep learning methods, and is robust against controller hyperparameter variation, demonstrated on domain generalization tasks with multi-dimensional regularization losses.
Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 31 of its 65 references that are in the map. First posted on arXiv 20 Mar 2024.
Builds on
31 of 65 references.
References · 31
- 2015Domain-Adversarial Training of Neural Networkscited by 3,044
- 2019Invariant Risk Minimizationcited by 1,854
- 2017Decoupled Weight Decay Regularizationcited by 1,831
- 2020In Search of Lost Domain Generalizationcited by 1,054
- 2017Deeper, Broader and Artier Domain Generalizationcited by 986
- 2016InfoGAN: Interpretable Representation Learning by Information Maximizing Generative Adversarial Netscited by 860
- 2019On the Convergence of Adam and Beyondcited by 841
- 2018Multi-Task Learning as Multi-Objective Optimizationcited by 821
- 2017Learning to Generalize: Meta-Learning for Domain Generalizationcited by 763
- 2018Reinforcement Learning and Control as Probabilistic Inference: Tutorial and Reviewcited by 482
- 2019Domain Generalization by Solving Jigsaw Puzzlescited by 450
- 2016Ladder Variational Autoencoderscited by 318
19 more
- 2016Hierarchical Deep Reinforcement Learning: Integrating Temporal Abstraction and Intrinsic Motivationcited by 286
- 2019Pareto Multi-Task Learningcited by 224
- 2021Fishr: Invariant Gradient Variances for Out-of-Distribution Generalizationcited by 186
- 2019DIVA: Domain Invariant Variational Autoencoderscited by 115
- 2018Tunability: Importance of Hyperparameters of Machine Learning Algorithmscited by 99
- 2018Taming VAEscited by 85
- 2019Learning Hierarchical Priors in VAEscited by 53
- 2021Domain Adversarial Neural Networks for Domain Generalization: When It Works and How to Improvecited by 46
- 2021Scalable Pareto Front Approximation for Deep Multi-Objective Learningcited by 45
- 2022Pareto Invariant Risk Minimization: Towards Mitigating the Optimization Dilemma in Out-of-Distribution Generalizationcited by 38
- 2018Piecewise Strong Convexity of Neural Networkscited by 13
- 2021Optimization Algorithms as Robust Feedback Controllerscited by 9
- 2021Domain Invariant Adversarial Learningcited by 7
- 2019High Dimensional Restrictive Federated Model Selection with multi-objective Bayesian Optimization over shifted distributionscited by 3
- 2021Hierarchical Variational Auto-Encoding for Unsupervised Domain Generalizationcited by 3
- 2019Variational Resampling Based Assessment of Deep Neural Networks under Distribution Shiftcited by 2
- 2019Tutorial and Survey on Probabilistic Graphical Model and Variational Inference in Deep Reinforcement Learningcited by 2
- 2019Closing the Loop: Dynamic State Estimation and Feedback Optimization of Power Gridscited by 1
- 2022Model-Free Nonlinear Feedback Optimizationcited by 1
Built on by
Not cited yet.