A Primal Dual Formulation For Deep Learning With Constraints

By Yatin Nandwani, Abhishek Pathak, Mausam, Parag Singla. 2019

122 citations recorded by the source corpus.

Subfields: Optimization & Training

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Optimization & Training

1. Simon Lacoste-Julien
2. Michael W. Mahoney
3. A. Gholami
4. Lu Lu
5. Yoshua Bengio
6. Jose Gallego-Posada
7. Francesc Verdugo
8. Zhicheng Wang

Citations in the corpus

• Controlled Sparsity via Constrained Optimization or: How I Learned to Stop Tuning Penalties and Love Constraints (2022, methodology): “Nandwani et al. [36] study the problem of training deep models under constraints on the network’s predicted labels, and approach the constrained problem in practice through a min-max Lagrangian formulation.”
• Characterizing possible failure modes in physics-informed neural networks (2021, background): “Previous work has tried to impose hard constraints in ML (both within the context of SciML and otherwise) [6, 21, 22, 24, 40], although this can be computationally expensive, and does not guarantee better results or convergence.”
• Physics-informed neural networks with hard constraints for inverse design (2021, background): “Im-posing hard constraints to neural networks has been considered very recently [36, 9] for data-driven problems in the supervised learning paradigm.”