Extending Adversarial Attacks and Defenses to Deep 3D Point Cloud Classifiers

@article{Liu2019ExtendingAA,
  title={Extending Adversarial Attacks and Defenses to Deep 3D Point Cloud Classifiers},
  author={Daniel Liu and Ronald Yu and Hao Su},
  journal={2019 IEEE International Conference on Image Processing (ICIP)},
  year={2019},
  pages={2279-2283}
}
3D object classification using deep neural networks has been extremely successful. As the problem of identifying 3D objects has many safety-critical applications, the neural networks have to be robust against adversarial changes to the input data set. We present a preliminary evaluation of adversarial attacks on 3D point cloud classifiers by evaluating adversarial attacks that were proposed for 2D images, and extending those attacks to reduce the perceptibility of the perturbations in 3D space… CONTINUE READING
4
Twitter Mentions

Figures, Tables, and Topics from this paper.

Citations

Publications citing this paper.
SHOWING 1-9 OF 9 CITATIONS

Adversarial point perturbations on 3D objects

VIEW 4 EXCERPTS
CITES BACKGROUND

Geometry-aware Generation of Adversarial and Cooperative Point Clouds

VIEW 4 EXCERPTS
CITES METHODS & BACKGROUND
HIGHLY INFLUENCED

Enhancing ML Robustness Using Physical-World Constraints

VIEW 2 EXCERPTS
CITES BACKGROUND

S4G: Amodal Single-view Single-Shot SE(3) Grasp Detection in Cluttered Scenes

VIEW 2 EXCERPTS
CITES METHODS & BACKGROUND

SampleNet: Differentiable Point Cloud Sampling

VIEW 1 EXCERPT
CITES BACKGROUND

DUP-Net: Denoiser and Upsampler Network for 3D Adversarial Point Clouds Defense.

VIEW 2 EXCERPTS
CITES METHODS & BACKGROUND

References

Publications referenced by this paper.
SHOWING 1-10 OF 27 REFERENCES

The Limitations of Deep Learning in Adversarial Settings

VIEW 6 EXCERPTS
HIGHLY INFLUENTIAL

Adversarial Machine Learning at Scale

VIEW 4 EXCERPTS
HIGHLY INFLUENTIAL

3D ShapeNets: A deep representation for volumetric shapes

VIEW 7 EXCERPTS
HIGHLY INFLUENTIAL

Explaining and Harnessing Adversarial Examples

VIEW 6 EXCERPTS
HIGHLY INFLUENTIAL

PointNet: Deep Learning on Point Sets for 3D Classification and Segmentation

VIEW 7 EXCERPTS

Generating 3D Adversarial Point Clouds

VIEW 1 EXCERPT