Verifiability and Predictability: Interpreting Utilities of Network Architectures for Point Cloud Processing
arXiv:1911.09053
Abstract
In this paper, we diagnose deep neural networks for 3D point cloud processing to explore utilities of different intermediate-layer network architectures. We propose a number of hypotheses on the effects of specific intermediate-layer network architectures on the representation capacity of DNNs. In order to prove the hypotheses, we design five metrics to diagnose various types of DNNs from the following perspectives, information discarding, information concentration, rotation robustness, adversarial robustness, and neighborhood inconsistency. We conduct comparative studies based on such metrics to verify the hypotheses. We further use the verified hypotheses to revise intermediate-layer architectures of existing DNNs and improve their utilities. Experiments demonstrate the effectiveness of our method.
References in corpus (18)
- The information bottleneck method
- Opening the Black Box of Deep Neural Networks via Information
- Object Detectors Emerge in Deep Scene CNNs
- Similarity of Neural Network Representations Revisited
- PointSIFT: A SIFT-like Network Module for 3D Point Cloud Semantic Segmentation
- Learning how to explain neural networks: PatternNet and PatternAttribution
- Dynamic Graph CNN for Learning on Point Clouds
- SVCCA: Singular Vector Canonical Correlation Analysis for Deep Learning Dynamics and Interpretability
- Explainable Neural Networks based on Additive Index Models
- SOM-VAE: Interpretable Discrete Representation Learning on Time Series
- A Closer Look at Deep Learning Heuristics: Learning rate restarts, Warmup and Distillation
- Point2Sequence: Learning the Shape Representation of 3D Point Clouds with an Attention-based Sequence to Sequence Network
- Towards Interpretable Reinforcement Learning Using Attention Augmented Agents
- Knowledge Consistency between Neural Networks and Beyond
- A Unified Approach to Interpreting and Boosting Adversarial Transferability
- 3D-Rotation-Equivariant Quaternion Neural Networks
- PointCloud Saliency Maps
- Interpreting and Boosting Dropout from a Game-Theoretic View