0 citations0 references

Teaching Compositionality to CNNs

2017pp. 732–741

Citations Over TimeTop 25% of 2017 papers

Austin Stone, Huayan Wang, Michael J. Stark, Yi Liu, D. Scott Phoenix, Dileep George

Abstract

Convolutional neural networks (CNNs) have shown great success in computer vision, approaching human-level performance when trained for specific tasks via application-specific loss functions. In this paper, we propose a method for augmenting and training CNNs so that their learned features are compositional. It encourages networks to form representations that disentangle objects from their surroundings and from each other, thereby promoting better generalization. Our method is agnostic to the specific details of the underlying CNN to which it is applied and can in principle be used with any CNN. As we show in our experiments, the learned representations lead to feature activations that are more localized and improve performance over non-compositional baselines in object recognition tasks.

Related Papers

→ Deep Residual Learning for Image Recognition(2016)216,989 cited
→ Gradient-based learning applied to document recognition(1998)57,014 cited
beta-VAE: Learning Basic Visual Concepts with a Constrained Variational Framework(2017)
→ Words as a window: Using word embeddings to explore the learned representations of Convolutional Neural Networks(2021)3 cited
→ Harnessing Deep Neural Networks with Logic Rules(2016)200 cited