ICLR 2018 accepted papers on Optimization
25 of the 337 papers accepted at ICLR 2018 (7.4%) are on Optimization, compared with 5.1% at ICLR 2017. They fall into 4 topics.
See ICLR 2018 in the interactive map
Submitting to ICLR? Checklist →
Working on something in this area? Describe your idea in scime atlas to see the closest papers by meaning and how crowded the spot has become.
Topics and papers
Optimizers · 13 papers
- On the Convergence of Adam and Beyond (1,616 citations)
- Don't Decay the Learning Rate, Increase the Batch Size (590 citations)
- Learning Sparse Neural Networks through L_0 Regularization (148 citations)
- Meta-Learning and Universality: Deep Representations and Gradient Descent can Approximate any Learning Algorithm (90 citations)
- On the insufficiency of existing momentum schemes for Stochastic Optimization (87 citations)
- Online Learning Rate Adaptation with Hypergradient Descent (75 citations)
- Deep Learning with Logged Bandit Feedback (66 citations)
- Kronecker-factored Curvature Approximations for Recurrent Neural Networks (44 citations)
- Understanding Short-Horizon Bias in Stochastic Meta-Optimization (42 citations)
- Unbiased Online Recurrent Optimization (25 citations)
- Deep Learning as a Mixed Convex-Combinatorial Optimization Problem (17 citations)
- Proximal Backpropagation (8 citations)
and 1 more in the interactive map.
Implicit bias and generalization · 9 papers
- On the Information Bottleneck Theory of Deep Learning (515 citations)
- Stochastic gradient descent performs variational inference, converges to limit cycles for deep networks (219 citations)
- Learning One-hidden-layer Neural Networks with Landscape Design (103 citations)
- When is a Convolutional Filter Easy to Learn? (59 citations)
- SGD Learns Over-parameterized Networks that Provably Generalize on Linearly Separable Data (30 citations)
- A Bayesian Perspective on Generalization and Stochastic Gradient Descent (23 citations)
- The Implicit Bias of Gradient Descent on Separable Data (15 citations)
- Empirical Risk Landscape Analysis for Understanding Deep Neural Networks (11 citations)
- Stabilizing Adversarial Nets with Prediction Methods (10 citations)
Nonconvex and smooth optimization · 2 papers
- On the regularization of Wasserstein GANs (133 citations)
- Training GANs with Optimism (120 citations)
