TY - GEN
T1 - SplineCam
T2 - 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2023
AU - Humayun, Ahmed Imtiaz
AU - Balestriero, Randall
AU - Balakrishnan, Guha
AU - Baraniuk, Richard
N1 - Funding Information:
Humayun and Baraniuk were supported by NSF grants CCF-1911094, IIS-1838177, and IIS-1730574; ONR grants N00014-18-12571, N00014-20-1-2534, and MURI N00014-20-1-2787; AFOSR grant FA9550-22-1-0060; and a Vannevar Bush Faculty Fellowship, ONR grant N00014-18-1-2047.
Publisher Copyright:
© 2023 IEEE.
PY - 2023
Y1 - 2023
N2 - Current Deep Network (DN) visualization and inter-pretability methods rely heavily on data space visualizations such as scoring which dimensions of the data are responsible for their associated prediction or generating new data features or samples that best match a given DN unit or representation. In this paper, we go one step further by developing the first provably exact method for computing the geometry of a DN's mapping - including its decision boundary - over a specified region of the data space. By lever-aging the theory of Continuous Piece- Wise Linear (CPWL) spline DNs, SplineCam exactly computes a DN's geometry without resorting to approximations such as sampling or architecture simplification. SplineCam applies to any DN architecture based on CPWL activation nonlinearities, including (leaky) ReLU, absolute value, maxout, and max-pooling and can also be applied to regression DNs such as implicit neural representations. Beyond decision boundary visualization and characterization, SplineCam enables one to compare architectures, measure generalizability, and sample from the decision boundary on or off the data manifold. Project website: bit.ly/splinecam.
AB - Current Deep Network (DN) visualization and inter-pretability methods rely heavily on data space visualizations such as scoring which dimensions of the data are responsible for their associated prediction or generating new data features or samples that best match a given DN unit or representation. In this paper, we go one step further by developing the first provably exact method for computing the geometry of a DN's mapping - including its decision boundary - over a specified region of the data space. By lever-aging the theory of Continuous Piece- Wise Linear (CPWL) spline DNs, SplineCam exactly computes a DN's geometry without resorting to approximations such as sampling or architecture simplification. SplineCam applies to any DN architecture based on CPWL activation nonlinearities, including (leaky) ReLU, absolute value, maxout, and max-pooling and can also be applied to regression DNs such as implicit neural representations. Beyond decision boundary visualization and characterization, SplineCam enables one to compare architectures, measure generalizability, and sample from the decision boundary on or off the data manifold. Project website: bit.ly/splinecam.
KW - Computer vision theory
UR - http://www.scopus.com/inward/record.url?scp=85164618731&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85164618731&partnerID=8YFLogxK
U2 - 10.1109/CVPR52729.2023.00369
DO - 10.1109/CVPR52729.2023.00369
M3 - Conference contribution
AN - SCOPUS:85164618731
T3 - Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition
SP - 3789
EP - 3798
BT - Proceedings - 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2023
PB - Institute of Electrical and Electronics Engineers Inc.
Y2 - 18 June 2023 through 22 June 2023
ER -