AAAI 2019 accepted papers on Architectures and efficiency
77 of the 1,147 papers accepted at AAAI 2019 (6.7%) are on Architectures and efficiency, compared with 7.2% at AAAI 2018. They fall into 7 topics.
See AAAI 2019 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
RNNs and normalization · 39 papers
- Latent Multi-Task Architecture Learning (327 citations)
- Predicting Hurricane Trajectories Using a Recurrent Neural Network (170 citations)
- Non-Autoregressive Machine Translation with Auxiliary Regularization (138 citations)
- SNR: Sub-Network Routing for Flexible Parameter Sharing in Multi-Task Learning (131 citations)
- Non-Autoregressive Neural Machine Translation with Enhanced Decoder Input (117 citations)
- Compressing Recurrent Neural Networks with Tensor Ring for Action Recognition (108 citations)
- Learning Anytime Predictions in Neural Networks via Adaptive Loss Balancing (87 citations)
- Biologically Motivated Algorithms for Propagating Local Target Representations (61 citations)
- Regularizing Neural Machine Translation by Target-Bidirectional Agreement (61 citations)
- Towards Non-Saturating Recurrent Units for Modelling Long-Term Dependencies (57 citations)
- PhoneMD: Learning to Diagnose Parkinson’s Disease from Smartphone Data (51 citations)
- DTMT: A Novel Deep Transition Architecture for Neural Machine Translation (48 citations)
and 27 more in the interactive map.
Attention and transformers · 18 papers
- Character-Level Language Modeling with Deeper Self-Attention (321 citations)
- Context-Aware Self-Attention Networks (126 citations)
- Recurrent Attention Model for Pedestrian Attribute Recognition (86 citations)
- Tied Transformers: Neural Machine Translation with Shared Encoder and Decoder (63 citations)
- Dynamic Layer Aggregation for Neural Machine Translation with Routing-by-Agreement (45 citations)
- Number Sequence Prediction Problems for Evaluating Computational Powers of Neural Networks (11 citations)
- A Grammar-Based Structural CNN Decoder for Code Generation
- A Task in a Suit and a Tie: Paraphrase Generation with Semantic Augmentation
- Fast and Simple Mixture of Softmaxes with BPE and Hybrid-LightRNN for Language Generation
- Gaussian Transformer: A Lightweight Approach for Natural Language Inference
- Generating Live Soccer-Match Commentary from Play Data
- Jointly Learning to Label Sentences and Tokens
and 6 more in the interactive map.
CNN design · 6 papers
- CNN-Cert: An Efficient Framework for Certifying Robustness of Convolutional Neural Networks (138 citations)
- Projection Convolutional Neural Networks for 1-bit CNNs via Discrete Back Propagation (102 citations)
- Weighted Channel Dropout for Regularization of Deep Convolutional Neural Network (85 citations)
- Learning to Steer by Mimicking Features from Heterogeneous Auxiliary Networks (67 citations)
- Super Sparse Convolutional Neural Networks (55 citations)
- CircConv: A Structured Convolution with Low Complexity (19 citations)
Quantization · 5 papers
- Deep Neural Network Quantization via Layer-Wise Optimization Using Limited Training Data (36 citations)
- Efficient Quantization for Neural Networks with Binary Weights and Low Bitwidth Activations (26 citations)
- Universal Approximation Property and Equivalence of Stochastic Computing-Based Neural Networks and Binary Neural Networks (13 citations)
- Multi-Precision Quantized Neural Networks via Encoding Decomposition of {-1,+1} (10 citations)
- Online Embedding Compression for Text Classification Using Low Rank Matrix Factorization
Pruning and efficient tuning · 4 papers
- Balanced Sparsity for Efficient DNN Inference on GPU (93 citations)
- Backbone Cannot Be Trained at Once: Rolling Back to Pre-Trained Network for Person Re-Identification (20 citations)
- Data Fine-Tuning (15 citations)
- A Layer Decomposition-Recomposition Framework for Neuron Pruning towards Accurate Lightweight Networks (13 citations)
