ICML 2017 accepted papers on Architectures and efficiency
47 of the 435 papers accepted at ICML 2017 (10.8%) are on Architectures and efficiency, compared with 8.1% at ICML 2016. They fall into 5 topics.
See ICML 2017 in the interactive map
Submitting to ICML? 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
RNNs and normalization · 35 papers
- Convolutional Sequence to Sequence Learning (2,604 citations)
- On Calibration of Modern Neural Networks (2,519 citations)
- Learning Important Features Through Propagating Activation Differences (2,386 citations)
- Large-Scale Evolution of Image Classifiers (534 citations)
- Deep Voice: Real-time Neural Text-to-Speech (393 citations)
- Device Placement Optimization with Reinforcement Learning (312 citations)
- Input Convex Neural Networks (191 citations)
- The Predictron: End-To-End Learning and Planning (150 citations)
- Tensor-Train Recurrent Neural Networks for Video Classification (142 citations)
- Efficient softmax approximation for GPUs (137 citations)
- OptNet: Differentiable Optimization as a Layer in Neural Networks (137 citations)
- Resource-efficient Machine Learning in 2 KB RAM for the Internet of Things (137 citations)
and 23 more in the interactive map.
CNN design · 8 papers
- Variational Dropout Sparsifies Deep Neural Networks (331 citations)
- Accelerating Eulerian Fluid Simulation With Convolutional Networks (289 citations)
- Equivariance Through Parameter-Sharing (112 citations)
- SplitNet: Learning to Semantically Split Deep Networks for Parameter Reduction and Model Parallelization (55 citations)
- Beyond Filters: Compact Feature Map for Portable Deep Model (37 citations)
- Convexified Convolutional Neural Networks (31 citations)
- Warped Convolutions: Efficient Invariance to Spatial Transformations (23 citations)
- Deep Tensor Convolution on Multicores (7 citations)
Attention and transformers · 2 papers
- Language Modeling with Gated Convolutional Networks (1,114 citations)
- Online and Linear-Time Attention by Enforcing Monotonic Alignments (225 citations)
Pruning and efficient tuning · 1 paper
- Combined Group and Exclusive Sparsity for Deep Neural Networks (116 citations)
