Hyperdimensional Feature Fusion for Out-Of-Distribution Detection
Wilson, Samuel, Sünderhauf, Niko, Dayoub, Feras
–arXiv.org Artificial Intelligence
We introduce powerful ideas from Hyperdimensional Computing into the challenging field of Out-of-Distribution (OOD) detection. In contrast to most existing work that performs OOD detection based on only a single layer of a neural network, we use similarity-preserving semi-orthogonal projection matrices to project the feature maps from multiple layers into a common vector space. By repeatedly applying the bundling operation $\oplus$, we create expressive class-specific descriptor vectors for all in-distribution classes. At test time, a simple and efficient cosine similarity calculation between descriptor vectors consistently identifies OOD samples with better performance than the current state-of-the-art. We show that the hyperdimensional fusion of multiple network layers is critical to achieve best general performance.
arXiv.org Artificial Intelligence
Dec-10-2021
- Country:
- North America
- Canada > Ontario (0.14)
- United States (0.28)
- North America
- Genre:
- Research Report (1.00)
- Technology: