AAAI 2024 accepted papers on Architectures and efficiency
152 of the 2,864 papers accepted at AAAI 2024 (5.3%) are on Architectures and efficiency, compared with 6.3% at AAAI 2023. They fall into 8 topics.
See AAAI 2024 in the interactive map
Working on something in this area? Describe your idea in scime atlas to see the closest accepted papers by meaning and how crowded the spot has become.
Topics and papers
RNNs and normalization · 46 papers
- GLOP: Learning Global Partition and Local Construction for Solving Large-Scale Routing Problems in Real-Time (38 citations)
- ModWaveMLP: MLP-Based Mode Decomposition and Wavelet Denoising Model to Defeat Complex Structures in Traffic Forecasting (24 citations)
- DeepSpeed Data Efficiency: Improving Deep Learning Model Quality and Training Efficiency via Efficient Data Sampling and Routing (23 citations)
- Recurrent Partial Kernel Network for Efficient Optical Flow Estimation (22 citations)
- Distilling Autoregressive Models to Obtain High-Performance Non-autoregressive Solvers for Vehicle Routing Problems with Faster Inference Speed (15 citations)
- Model-Driven Deep Neural Network for Enhanced AoA Estimation Using 5G gNB (12 citations)
- Learning Temporal Resolution in Spectrogram for Audio Classification (11 citations)
- Inducing Point Operator Transformer: A Flexible and Scalable Architecture for Solving PDEs (10 citations)
- Grab What You Need: Rethinking Complex Table Structure Recognition with Flexible Components Deliberation (9 citations)
- A Model for Estimating the Economic Costs of Computer Vision Systems That Use Deep Learning (8 citations)
- End-to-End Learning of LTLf Formulae by Faithful LTLf Encoding (8 citations)
- Layer Collaboration in the Forward-Forward Algorithm (6 citations)
and 34 more in the interactive map.
Attention and transformers · 44 papers
- IRPruneDet: Efficient Infrared Small Target Detection via Wavelet Structure-Regularized Soft Channel Pruning (155 citations)
- An Attentive Inductive Bias for Sequential Recommendation beyond the Self-Attention (124 citations)
- FFT-Based Dynamic Token Mixer for Vision (64 citations)
- S2WAT: Image Style Transfer via Hierarchical Vision Transformer Using Strips Window Attention (50 citations)
- PMET: Precise Model Editing in a Transformer (37 citations)
- Vision Transformer Off-the-Shelf: A Surprising Baseline for Few-Shot Class-Agnostic Counting (29 citations)
- Behavioral Recognition of Skeletal Data Based on Targeted Dual Fusion Strategy (18 citations)
- Rethinking Attention: Exploring Shallow Feed-Forward Neural Networks as an Alternative to Attention Layers in Transformers (Student Abstract) (17 citations)
- Bi-ViT: Pushing the Limit of Vision Transformer Quantization (16 citations)
- ViTEraser: Harnessing the Power of Vision Transformers for Scene Text Removal with SegMIM Pretraining (16 citations)
- Dual-Window Multiscale Transformer for Hyperspectral Snapshot Compressive Imaging (15 citations)
- LF-ViT: Reducing Spatial Redundancy in Vision Transformer for Efficient Image Recognition (15 citations)
and 32 more in the interactive map.
CNN design · 18 papers
- Rolling-Unet: Revitalizing MLP’s Ability to Efficiently Extract Long-Distance Dependencies for Medical Image Segmentation (138 citations)
- Convolutional Channel-Wise Competitive Learning for the Forward-Forward Algorithm (12 citations)
- Another Way to the Top: Exploit Contextual Clustering in Learned Image Coding (11 citations)
- Interpretability Benchmark for Evaluating Spatial Misalignment of Prototypical Parts Explanations (10 citations)
- DTF-AT: Decoupled Time-Frequency Audio Transformer for Event Classification (6 citations)
- Multiscale Low-Frequency Memory Network for Improved Feature Extraction in Convolutional Neural Networks (6 citations)
- PICNN: A Pathway towards Interpretable Convolutional Neural Networks (6 citations)
- REPrune: Channel Pruning via Kernel Representative Selection (6 citations)
- Efficient Look-Up Table from Expanded Convolutional Network for Accelerating Image Super-resolution (5 citations)
- NeSyFOLD: A Framework for Interpretable Image Classification (5 citations)
- Block Image Compressive Sensing with Local and Global Information Interaction (3 citations)
- RLPeri: Accelerating Visual Perimetry Test with Reinforcement Learning and Convolutional Feature Extraction (3 citations)
and 6 more in the interactive map.
Pruning and efficient tuning · 16 papers
- VMT-Adapter: Parameter-Efficient Transfer Learning for Multi-Task Dense Scene Understanding (62 citations)
- SAM-PARSER: Fine-Tuning SAM Efficiently by Parameter Space Reconstruction (28 citations)
- LION: Implicit Vision Prompt Tuning (26 citations)
- DTL: Disentangled Transfer Learning for Visual Recognition (21 citations)
- AdapterGNN: Parameter-Efficient Fine-Tuning Improves Generalization in GNNs (19 citations)
- UPDP: A Unified Progressive Depth Pruner for CNN and Vision Transformer (17 citations)
- Data Shunt: Collaboration of Small and Large Models for Lower Costs and Better Performance (15 citations)
- HyperFast: Instant Classification for Tabular Data (15 citations)
- Entropy Induced Pruning Framework for Convolutional Neural Networks (8 citations)
- EPSD: Early Pruning with Self-Distillation for Efficient Model Compression (7 citations)
- Fast and Controllable Post-training Sparsity: Learning Optimal Sparsity Allocation with Global Constraint in Minutes (7 citations)
- Some Like It Small: Czech Semantic Embedding Models for Industry Applications (4 citations)
and 4 more in the interactive map.
Spiking networks · 10 papers
- SpikingBERT: Distilling BERT to Train Spiking Language Models Using Implicit Differentiation (46 citations)
- Gated Attention Coding for Training High-Performance and Efficient Spiking Neural Networks (44 citations)
- Ternary Spike: Learning Ternary Spikes for Spiking Neural Networks (40 citations)
- Shrinking Your TimeStep: Towards Low-Latency Neuromorphic Object Recognition with Spiking Neural Networks (28 citations)
- TC-LIF: A Two-Compartment Spiking Neuron Model for Long-Term Sequential Modelling (26 citations)
- Memory-Efficient Reversible Spiking Neural Networks (13 citations)
- Enhancing Representation of Spiking Neural Networks via Similarity-Sensitive Contrastive Learning (11 citations)
- Enhancing Training of Spiking Neural Network with Stochastic Latency (11 citations)
- Spiking NeRF: Representing the Real-World Geometry by a Discontinuous Representation (10 citations)
- Finding Visual Saliency in Continuous Spike Stream (6 citations)
Quantization · 9 papers
- OWQ: Outlier-Aware Weight Quantization for Efficient Fine-Tuning and Inference of Large Language Models (72 citations)
- Agile-Quant: Activation-Guided Quantization for Faster Inference of LLMs on the Edge (35 citations)
- Make RepVGG Greater Again: A Quantization-Aware Approach (33 citations)
- Exploring Post-training Quantization in LLMs from Comprehensive Study to Low Rank Compensation (22 citations)
- Towards Efficient Verification of Quantized Neural Networks (17 citations)
- Norm Tweaking: High-Performance Low-Bit Quantization of Large Language Models (10 citations)
- PTMQ: Post-training Multi-Bit Quantization of Neural Networks (10 citations)
- AQ-DETR: Low-Bit Quantized Detection Transformer with Auxiliary Queries (8 citations)
- MetaMix: Meta-State Precision Searcher for Mixed-Precision Activation Quantization (4 citations)
Neural architecture search · 7 papers
- Training-Free Quantum Architecture Search (26 citations)
- SasWOT: Real-Time Semantic Segmentation Architecture Search WithOut Training (23 citations)
- DC-NAS: Divide-and-Conquer Neural Architecture Search for Multi-Modal Classification (21 citations)
- IS-DARTS: Stabilizing DARTS through Precise Measurement on Candidate Importance (20 citations)
- Auto-Prox: Training-Free Vision Transformer Architecture Search via Automatic Proxy Discovery (19 citations)
- UniADS: Universal Architecture-Distiller Search for Distillation Gap (16 citations)
- Hypergraph Neural Architecture Search (4 citations)