Deep Set Prediction Networks

Abstract

Current approaches for predicting sets from feature vectors ignore the unordered nature of sets and suffer from discontinuity issues as a result. We propose a general model for predicting sets that properly respects the structure of sets and avoids this problem. With a single feature vector as input, we show that our model is able to auto-encode point sets, predict the set of bounding boxes of objects in an image, and predict the set of attributes of these objects.

Appendix C contains an erratum

References in corpus (10)

• Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks
• Learning Representations and Generative Models for 3D Point Clouds
• GraphRNN: Generating Realistic Graphs with Deep Auto-regressive Models
• MolGAN: An implicit generative model for small molecular graphs
• Learning Deep Generative Models of Graphs
• Multi-Object Representation Learning with Iterative Variational Inference
• Learning Latent Permutations with Gumbel-Sinkhorn Networks
• FSPool: Learning Set Representations with Featurewise Sort Pooling
• Deep Perm-Set Net: Learn to predict sets with unknown permutation and cardinality using deep neural networks
• Concept Learning with Energy-Based Models

Cited by in corpus (22)

• Object-Centric Learning with Slot Attention
• The Dark Machines Anomaly Score Challenge: Benchmark Data and Model Independent Event Classification for the Large Hadron Collider
• A Survey on Influence Maximization: From an ML-Based Combinatorial Optimization
• FSPool: Learning Set Representations with Featurewise Sort Pooling
• Deep Set Auto Encoders for Anomaly Detection in Particle Physics
• Entity Abstraction in Visual Model-Based Reinforcement Learning
• Tackling the Challenges in Scene Graph Generation with Local-to-Global Interactions
• Hierarchical Poset Decoding for Compositional Generalization in Language
• GENESIS-V2: Inferring Unordered Object Representations without Iterative Refinement
• The Differentiable Cross-Entropy Method
• Visual Representation Learning Does Not Generalize Strongly Within the Same Domain
• Set-to-Sequence Methods in Machine Learning: a Review
• Non-autoregressive Personalized Bundle Generation
• Better Set Representations For Relational Reasoning
• Learning to Manipulate Object Collections Using Grounded State Representations
• SLASH: Embracing Probabilistic Circuits into Neural Answer Set Programming
• A deep learning guided memetic framework for graph coloring problems
• Multiset-Equivariant Set Prediction with Approximate Implicit Differentiation
• Deep set conditioned latent representations for action recognition
• USCO-Solver: Solving Undetermined Stochastic Combinatorial Optimization Problems
• Top-N: Equivariant set and graph generation without exchangeability
• Set Prediction without Imposing Structure as Conditional Density Estimation