DRAW: A Recurrent Neural Network For Image Generation

2015/02/16 by Karol Gregor, Ivo Danihelka, Gregor, Karol +7 · 2 voices · 80 citations
Computer Science · #Generative Adversarial Networks and Image Synthesis #Image Retrieval and Classification Techniques #Medical Image Segmentation Techniques #cs.CV #cs.LG #cs.NE

paper · pdf · doi:10.48550/arxiv.1502.04623

openalex publication_date 2015/02/16 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

This paper introduces the Deep Recurrent Attentive Writer (DRAW) neural network architecture for image generation. DRAW networks combine a novel spatial attention mechanism that mimics the foveation of the human eye, with a sequential variational auto-encoding framework that allows for the iterative construction of complex images. The system substantially improves on the state of the art for generative models on MNIST, and, when trained on the Street View House Numbers dataset, it generates images that cannot be distinguished from real data with the naked eye.

Citations

• Auto-Encoding Variational Bayes
• Gradient-based learning applied to document recognition
• Learning Multiple Layers of Features from Tiny Images

Cited by

• Mechanisms of Projective Composition of Diffusion Models
• Ksurf-Drone: Attention Kalman Filter for Contextual Bandit Optimization in Cloud Resource Allocation
• Neural Machine Translation and Sequence-to-sequence Models: A Tutorial
• Composing graphical models with neural networks for structured representations and fast inference
• Spherical Latent Spaces for Stable Variational Autoencoders
• Attention-GAN for Object Transfiguration in Wild Images
• Inverse Reinforcement Learning via Deep Gaussian Process
• Nonparametric Variational Auto-encoders for Hierarchical Representation Learning
• Hierarchical Deep Reinforcement Learning: Integrating Temporal Abstraction and Intrinsic Motivation
• Generative AI in Depth: A Survey of Recent Advances, Model Variants, and Real-World Applications
• Count-Based Exploration with Neural Density Models
• GP-GAN: Towards Realistic High-Resolution Image Blending
• Text-guided Attention Model for Image Captioning
• Compositional Visual Generation and Inference with Energy Based Models
• Tutorial on Variational Autoencoders
• Deep Feature Consistent Variational Autoencoder
• Layer Normalization
• Visual7W: Grounded Question Answering in Images
• Building machines that learn and think like people
• Bounded Information Rate Variational Autoencoders
• Emergent Symbols through Binding in External Memory
• Designing ReLU Generative Networks to Enumerate Trees with a Given Tree Edit Distance
• Artificial Intelligence in Quantitative Ultrasound Imaging: A Review
• Face Parsing via Recurrent Propagation
• Computational Graph Approach for Detection of Composite Human Activities
• Learning to Decompose and Disentangle Representations for Video Prediction
• Text-to-Image Generation with Attention Based Recurrent Neural Networks
• Recurrent Spatial Transformer Networks
• Harnessing Synthetic Data from Generative AI for Statistical Inference
• Connectionist Recommendation in the Wild: On the utility and scrutability of neural networks for personalized course guidance
• A Generative Learning Approach for Spatio-temporal Modeling in Connected Vehicular Network
• Iterative Retraining of Quantum Spin Models Using Recurrent Neural Networks
• Deep Nonparametric Estimation of Discrete Conditional Distributions via Smoothed Dyadic Partitioning
• Dynamic Deep Neural Networks: Optimizing Accuracy-Efficiency Trade-offs by Selective Execution
• Depth Structure Preserving Scene Image Generation
• From Language Games to Drawing Games
• An overview of neural architectures for self-supervised audio representation learning from masked spectrograms
• Unsupervised Learning of 3D Structure from Images
• Variational inference for Monte Carlo objectives
• Learning FRAME Models Using CNN Filters
• A Fully Trainable Network with RNN-based Pooling
• Variable Rate Image Compression with Recurrent Neural Networks
• Variational Walkback: Learning a Transition Operator as a Stochastic\n Recurrent Net
• Improving Clinical Predictions through Unsupervised Time Series Representation Learning
• Semi-Implicit Variational Inference
• Sparsity in Variational Autoencoders
• Deep, Convolutional, and Recurrent Models for Human Activity Recognition using Wearables
• Bio-inspired visual attention for silicon retinas based on spiking neural networks applied to pattern classification
• Blur-Attention: A boosting mechanism for non-uniform blurred image restoration
• Learning Residual Images for Face Attribute Manipulation
• AMTnet: Action-Micro-Tube Regression by End-to-end Trainable Deep\n Architecture
• Adaptive Computation Time for Recurrent Neural Networks
• Dual Attention Networks for Multimodal Reasoning and Matching
• Dirichlet Variational Autoencoder for Text Modeling
• Psychlab: A Psychology Laboratory for Deep Reinforcement Learning Agents
• Deep Recurrent Generative Decoder for Abstractive Text Summarization
• Parametrized Accelerated Methods Free of Condition Number
• AttnGAN: Fine-Grained Text to Image Generation with Attentional Generative Adversarial Networks
• Understanding in Artificial Intelligence
• Every Smile is Unique: Landmark-Guided Diverse Smile Generation
• Unveiling the Dreams of Word Embeddings: Towards Language-Driven Image Generation
• Spatial Broadcast Decoder: A Simple Architecture for Learning Disentangled Representations in VAEs
• CDVAE: Co-embedding Deep Variational Auto Encoder for Conditional Variational Generation
• Variational Rejection Sampling
• Z-Forcing: Training Stochastic Recurrent Networks
• AttZoom: Attention Zoom for Better Visual Features
• Modeling documents with Generative Adversarial Networks
• Teaching Machines to Read and Comprehend
• Generative Mixture of Networks
• Variational Bi-LSTMs
• Hierarchical Attentive Recurrent Tracking
• Real-time interactive sequence generation and control with Recurrent\n Neural Network ensembles
• Think Visually: Question Answering through Virtual Imagery
• One-element Batch Training by Moving Window
• Sampled Training and Node Inheritance for Fast Evolutionary Neural Architecture Search
• Data Augmentation for Enhancing EEG-based Emotion Recognition with Deep Generative Models
• Encoding Spatial Relations from Natural Language
• Convergence guarantees for RMSProp and ADAM in non-convex optimization and an empirical comparison to Nesterov acceleration
• Endowing Deep 3D Models with Rotation Invariance Based on Principal Component Analysis
• On Generalization Bounds of a Family of Recurrent Neural Networks

Discussions

• DRAW: A Recurrent Neural Network For Image Generation [pdf] [hn, 26 points, 7 comments]
• DRAW: A Recurrent Neural Network for Image Generation [pdf] [hn, 2 points, 0 comments]

Related

• A Neural Representation of Sketch Drawings
• Sketch-pix2seq: a Model to Generate Sketches of Multiple Categories
• Generate Image Descriptions based on Deep RNN and Memory Cells for Images Features
• Enabling My Robot To Play Pictionary : Recurrent Neural Networks For Sketch Recognition
• Generative AI for Vision: A Comprehensive Study of Frameworks and Applications