ECCV 2020 accepted papers on Architectures and efficiency
86 of the 1,358 papers accepted at ECCV 2020 (6.3%) are on Architectures and efficiency, compared with 5.9% at ECCV 2018. They fall into 7 topics.
See ECCV 2020 in the interactive map
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
Neural architecture search · 26 papers
- Single Path One-Shot Neural Architecture Search with Uniform Sampling (930 citations)
- Fair DARTS: Eliminating Unfair Advantages in Differentiable Architecture Search (357 citations)
- BigNAS: Scaling Up Neural Architecture Search with Big Single-Stage Models (256 citations)
- Faster AutoAugment: Learning Augmentation Strategies Using Backpropagation (210 citations)
- Neural Predictor for Neural Architecture Search (200 citations)
- DHP: Differentiable Meta Pruning via HyperNetworks (137 citations)
- NAS-Count: Counting-by-Density with Neural Architecture Search (119 citations)
- A Generic Graph-based Neural Architecture Encoding Scheme for Predictor-based NAS (100 citations)
- Are Labels Necessary for Neural Architecture Search? (73 citations)
- BATS: Binary ArchitecTure Search (68 citations)
- Angle-based Search Space Shrinking for Neural Architecture Search (64 citations)
- Off-Policy Reinforcement Learning for Efficient and Effective GAN Architecture Search (63 citations)
and 14 more in the interactive map.
CNN design · 20 papers
- Efficient Neighbourhood Consensus Networks via Submanifold Sparse Convolutions (189 citations)
- Adaptive Mixture Regression Network with Local Counting Map for Crowd Counting (149 citations)
- WeightNet: Revisiting the Design Space of Weight Networks (108 citations)
- Context-Gated Convolution (72 citations)
- Dynamic Group Convolution for Accelerating Convolutional Neural Networks (64 citations)
- Fully Embedding Fast Convolutional Networks on Pixel Processor Arrays (43 citations)
- PSConv: Squeezing Feature Pyramid into One Compact Poly-Scale Convolutional Layer (39 citations)
- LST-Net: Learning a Convolutional Neural Network with a Learnable Sparse Transform (32 citations)
- Thanks for Nothing: Predicting Zero-Valued Activations with Lightweight Convolutional Neural Networks (24 citations)
- Efficient Residue Number System Based Winograd Convolution (15 citations)
- MABNet: A Lightweight Stereo Network Based on Multibranch Adjustable Bottleneck Module (14 citations)
- Learnable Cost Volume Using the Cayley Representation (12 citations)
and 8 more in the interactive map.
RNNs and normalization · 14 papers
- Rethinking Bottleneck Structure for Efficient Mobile Network Design (304 citations)
- Dynamic ReLU (194 citations)
- Forgetting Outside the Box: Scrubbing Deep Networks of Information Accessible from Input-Output Observations (146 citations)
- Channel selection using Gumbel Softmax (69 citations)
- Momentum Batch Normalization for Deep Learning with Small Batch Size (48 citations)
- Resolution Switchable Networks for Runtime Efficient Image Recognition (33 citations)
- Toward Faster and Simpler Matrix Normalization via Rank-1 Update (18 citations)
- HALO: Hardware-Aware Learning to Optimize (16 citations)
- SemifreddoNets: Partially Frozen Neural Networks for Efficient Computer Vision Systems (14 citations)
- Learning Propagation Rules for Attribution Map Generation (9 citations)
- Accelerating Deep Learning with Millions of Classes (3 citations)
- An Efficient Training Framework for Reversible Neural Architectures (1 citation)
and 2 more in the interactive map.
Quantization · 10 papers
- ReActNet: Towards Precise Binary Neural Network with Generalized Activation Functions (408 citations)
- PAMS: Quantized Super-Resolution via Parameterized Max Scale (91 citations)
- Differentiable Joint Pruning and Quantization for Hardware Efficiency (83 citations)
- PROFIT: A Novel Training Method for sub-4-bit MobileNet Models (76 citations)
- HMQ: Hardware Friendly Mixed Precision Quantization Block for CNNs (58 citations)
- ProxyBNN: Learning Binarized Neural Networks via Proxy Matrices (26 citations)
- Deep Transferring Quantization (18 citations)
- DBQ: A Differentiable Branch Quantizer for Lightweight Deep Neural Networks (8 citations)
- Finding Non-Uniform Quantization Schemes using Multi-Task Gaussian Processes (5 citations)
- Post-Training Piecewise Linear Quantization for Deep Neural Networks
Pruning and efficient tuning · 9 papers
- Big Transfer (BiT): General Visual Representation Learning (1,221 citations)
- Side-Tuning: A Baseline for Network Adaptation via Additive Side Networks (216 citations)
- Stable Low-rank Tensor Decomposition for Compression of Convolutional Neural Network (150 citations)
- DSA: More Efficient Budgeted Pruning via Differentiable Sparsity Allocation (105 citations)
- Accelerating CNN Training by Pruning Activation Gradients (40 citations)
- An Image Enhancing Pattern-based Sparsity for Real-time Inference on Mobile Devices (34 citations)
- A Generic Visualization Approach for Convolutional Neural Networks (9 citations)
- EagleEye: Fast Sub-net Evaluation for Efficient Neural Network Pruning
- Mapillary Planet-Scale Depth Dataset
Attention and transformers · 6 papers
- Axial-DeepLab: Stand-Alone Axial-Attention for Panoptic Segmentation (850 citations)
- End-to-End Low Cost Compressive Spectral Imaging with Spatial-Spectral Self-Attention (270 citations)
- Efficient Attention Mechanism for Visual Dialog that can Handle All the Interactions between Multiple Inputs (55 citations)
- Attention-Based Query Expansion Learning (41 citations)
- Weight Excitation: Built-in Attention Mechanisms in Convolutional Neural Networks (30 citations)
- Deep Reinforced Attention Learning for Quality-Aware Visual Recognition (9 citations)
