Abstract
The training of deep neural network classifiers results in decision boundaries whose geometry is still not well understood. This is in direct relation with classification problems such as so called corner case inputs. We introduce zoNNscan, an index that is intended to inform on the boundary uncertainty (in terms of the presence of other classes) around one given input datapoint. It is based on confidence entropy, and is implemented through Monte Carlo sampling in the multidimensional ball surrounding that input. We detail the zoNNscan index, give an algorithm for approximating it, and finally illustrate its benefits on three applications.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
For more advanced sampling techniques in high dimensional spaces, please refer to e.g., [5].
- 2.
Note that for \(X\in [0,1]^d\) and for \(r\ge 1\), \(\mathbf {B}_{\boldsymbol{\infty } }(X,r)\cap [0,1]^d = [0,1]^d\), then the space of normalized data is totally covered by the sampling process.
References
Keras: Deep learning for humans. https://github.com/keras-team/keras/tree/master/examples. Accessed 25 Feb 2020
Plotting high-dimensional decision boundaries. https://github.com/tmadl/highdimensional-decision-boundary-plot. Accessed 1 July 2018
Belongie, S., Malik, J., Puzicha, J.: Shape matching and object recognition using shape contexts. IEEE Trans. Pattern Anal. Mach. Intell. 24 (2002). https://doi.org/10.1109/34.993558
van den Berg, E.: Some insights into the geometry and training of neural networks. CoRR arXiv:1605.00329 (2016)
Dick, J., Kuo, F.Y., Sloan, I.H.: High-dimensional integration: the quasi-monte carlo way. Acta Numer. 22, 133–288 (2013). https://doi.org/10.1017/S0962492913000044
Fawzi, A., Moosavi-Dezfooli, S., Frossard, P., Soatto, S.: Classification regions of deep neural networks. CoRR arXiv:1705.09552 (2017)
Goodfellow, I.J., Shlens, J., Szegedy, C.: Explaining and harnessing adversarial examples. In: ICLR (2015)
Guo, G., Zhang, H.J., Li, S.Z.: Distance-from-boundary as a metric for texture image retrieval. In: IEEE International Conference on Acoustics, Speech, and Signal Processing, vol. 3, pp. 1629–1632 (2001). https://doi.org/10.1109/ICASSP.2001.941248
Kuncheva, L., Whitaker, C., Shipp, C., Duin, R.: Limits on the majority vote accuracy in classifier fusion. Pattern Anal. Appl. 6(1), 22–31 (2003)
Lakshminarayanan, B., Pritzel, A., Blundell, C.: Simple and scalable predictive uncertainty estimation using deep ensembles. In: Guyon, I., et al. (eds.) Advances in Neural Information Processing Systems 30, pp. 6402–6413. Curran Associates, Inc. (2017). http://papers.nips.cc/paper/7219-simple-and-scalable-predictive-uncertainty-estimation-using-deep-ensembles.pdf
Le Merrer, E., Pérez, P., Trédan, G.: Adversarial frontier stitching for remote neural network watermarking. Neural Comput. Appl. 32(13), 9233–9244 (2019). https://doi.org/10.1007/s00521-019-04434-z
Lee, C., Landgrebe, D.A.: Feature extraction based on decision boundaries. IEEE Trans. Pattern Anal. Mach. Intell. 15(4), 388–400 (1993)
Lee, C., Landgrebe, D.A.: Decision boundary feature extraction for neural networks. IEEE Trans. Neural Netw. 8(1), 75–83 (1997). https://doi.org/10.1109/72.554193
Liu, Y., et al.: Trojaning attack on neural networks. In: NDSS (2018)
Mann, H.B., Whitney, D.R.: On a test of whether one of two random variables is stochastically larger than the other. Ann. Math. Stat. 18(1), 50–60 (1947). https://doi.org/10.1214/aoms/1177730491
Moosavi-Dezfooli, S., Fawzi, A., Fawzi, O., Frossard, P.: Universal adversarial perturbations. In: CVPR (2017)
Pei, K., Cao, Y., Yang, J., Jana, S.: Deepxplore: automated whitebox testing of deep learning systems. In: SOSP (2017)
Wang, Z., Bovik, A.C.: A universal image quality index. IEEE Signal Process. Lett. 9(3), 81–84 (2002). https://doi.org/10.1109/97.995823
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2021 Springer Nature Switzerland AG
About this paper
Cite this paper
Jaouen, A., Le Merrer, E. (2021). zoNNscan: A Boundary-Entropy Index for Zone Inspection of Neural Models. In: Cazenave, T., Teytaud, O., Winands, M.H.M. (eds) Monte Carlo Search. MCS 2020. Communications in Computer and Information Science, vol 1379. Springer, Cham. https://doi.org/10.1007/978-3-030-89453-5_3
Download citation
DOI: https://doi.org/10.1007/978-3-030-89453-5_3
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-89452-8
Online ISBN: 978-3-030-89453-5
eBook Packages: Computer ScienceComputer Science (R0)