Generative Neurosymbolic Machines
By Jindong Jiang, Sungjin Ahn. 2020
73 citations recorded by the source corpus.
Reconciling symbolic and distributed representations is a crucial challenge that can potentially resolve the limitations of current deep learning. Remarkable advances in this direction have been achieved recently via generative object-centric representation models. While learning a recognition model that infers object-centric symbolic representations like bounding boxes from raw images in an unsupervised way, no such model can provide another important ability of a generative model, i.e., generating (sampling) according to the structure of learned world density. In this paper, we propose Generative Neurosymbolic Machines, a generative model that combines the benefits of distributed and symbolic representations to support both structured representations of symbolic components and density-based generation. These two crucial properties are achieved by a two-layer latent hierarchy with the global distributed latent for flexible density modeling and the structured symbolic latent map. To increase the model flexibility in this hierarchical structure, we also propose the StructDRAW prior. In experiments, we show that the proposed model significantly outperforms the previous structured representation models as well as the state-of-the-art non-structured generative models in terms of both structure accuracy and image generation quality.
Subfields: Computer Vision and Pattern Recognition
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Computer Vision and Pattern Recognition
- Sjoerd van Steenkiste
- J. Urgen Schmidhuber
- Klaus Greff
- Luc Van Gool
- Andrew Zisserman
- Richard Bowden
- Jitendra Malik
- Ming-Hsuan Yang
Citations in the corpus
- On the Binding Problem in Artificial Neural Networks (2020, methodology): “, 2020), or parallel spatial (bounding-box) attention (Lin et al., 2020; Jiang and Ahn, 2020).”
- D’ya Like DAGs? A Survey on Structure Learning and Causal Discovery (2021, background): “However, such reasoning is only possible if one already has access to variables which represent high-level semantic concepts, which is not the case when learning from raw video data, for example.”