![](bg6.png)
8 CONCLUSION
In the paper we focus on DeepFake detection, that be-
comes a serious threat to the privacy and biometric
security. We put particular interest in providing meth-
ods of detecting DeepFakes feasible for consumer
grade devices allowing anyone to verify, with reason-
able probability, if the video is legitimate. Our con-
tribution is twofold – we investigated the efficacy of
a shallow neural network MesoNet, with various ac-
tivation functions. Aside experimental verification of
previously introduced function, we presented a novel
one – Pish, that achieves results competitive to top-
performing ones. We achieve over 1% increase over
the original solution in both average test accuracy and
maximal test accuracy with Swish and Pish activation
functions, respectively. Such increase is substantial,
as even short DeepFake video (1 minute in length)
consists of 1500 images. This allows the user to in-
crease his certainty about the authenticity of a video
on his own personal computer.
Moreover, we present an evaluation of perfor-
mance of Pish under some well known networks. The
tests show that the new activation function may be an
interesting alternative. Further research in the aspect
of activation function should focus on providing op-
timized implementation of its calculation, so that it
generates lower overhead, moreover evaluation of the
function on other networks may provide interesting
applications resulting in high accuracy.
ACKNOWLEDGEMENTS
This work is partially supported by Polish
National Science Centre – project UMO-
2018/29/B/ST6/02969.
REFERENCES
Afchar, D., Nozick, V., Yamagishi, J., and Echizen, I.
(2018). Mesonet: a compact facial video forgery de-
tection network. CoRR, abs/1809.00888.
Akhtar, Z. and Dasgupta, D. (2019). A comparative evalu-
ation of local feature descriptors for deepfakes detec-
tion. In 2019 IEEE International Symposium on Tech-
nologies for Homeland Security (HST), pages 1–5.
Chollet, F. (2016). Xception: Deep learning with depthwise
separable convolutions. CoRR, abs/1610.02357.
Dolhansky, B., Bitton, J., Pflaum, B., Lu, J., Howes, R.,
Wang, M., and Ferrer, C. C. (2020). The deepfake
detection challenge (dfdc) dataset.
Durall, R., Keuper, M., Pfreundt, F.-J., and Keuper, J.
(2019). Unmasking deepfakes with simple features.
Glorot, X., Bordes, A., and Bengio, Y. (2011). Deep sparse
rectifier neural networks. Proceedings of the 14th In-
ternational Conference on Artificial Intelligence and
Statisitics (AISTATS) 2011, 15:315–323.
Goodfellow, I. J., Pouget-Abadie, J., Mirza, M., Xu, B.,
Warde-Farley, D., Ozair, S., Courville, A., and Ben-
gio, Y. (2014). Generative adversarial networks.
He, K., Zhang, X., Ren, S., and Sun, J. (2015). Deep
residual learning for image recognition. CoRR,
abs/1512.03385.
Iandola, F. N., Moskewicz, M. W., Ashraf, K., Han, S.,
Dally, W. J., and Keutzer, K. (2016). Squeezenet:
Alexnet-level accuracy with 50x fewer parameters and
<1mb model size. CoRR, abs/1602.07360.
Krizhevsky, A. (2012). Learning multiple layers of features
from tiny images. University of Toronto.
Lecun, Y., Bottou, L., Bengio, Y., and Haffner, P. (1998).
Gradient-based learning applied to document recogni-
tion. Proceedings of the IEEE, 86(11):2278–2324.
Lyu, S. (2020). Deepfake detection: Current challenges and
next steps.
Maas, A. L., Hannun, A. Y., and Ng, A. Y. (2013). Rec-
tifier nonlinearities improve neural network acoustic
models. In in ICML Workshop on Deep Learning for
Audio, Speech and Language Processing.
Misra, D. (2019). Mish: A self regularized non-monotonic
neural activation function.
Misra, D. (2021). digantamisra98 mish github repository.
https://github.com/digantamisra98/Mish. Accessed:
15-03-2021.
Nguyen, T. T., Nguyen, C. M., Nguyen, D. T., Nguyen,
D. T., and Nahavandi, S. (2020). Deep learning for
deepfakes creation and detection: A survey.
Ramachandran, P., Zoph, B., and Le, Q. V. (2017). Search-
ing for activation functions.
R
¨
ossler, A., Cozzolino, D., Verdoliva, L., Riess, C., Thies,
J., and Nießner, M. (2019). FaceForensics++: Learn-
ing to detect manipulated facial images. In Interna-
tional Conference on Computer Vision (ICCV).
Seferbekov, S. (2020). Kaggle dfdc best solution. https:
//www.kaggle.com/c/deepfake-detection-challenge/
discussion/145721. Accessed: 08-03-2021.
Szegedy, C., Liu, W., Jia, Y., Sermanet, P., Reed, S. E.,
Anguelov, D., Erhan, D., Vanhoucke, V., and Rabi-
novich, A. (2014). Going deeper with convolutions.
CoRR, abs/1409.4842.
Tan, M. and Le, Q. V. (2019). Efficientnet: Rethinking
model scaling for convolutional neural networks.
Tolosana, R., Vera-Rodriguez, R., Fierrez, J., Morales, A.,
and Ortega-Garcia, J. (2020). Deepfakes and beyond:
A survey of face manipulation and fake detection.
Yang, X., Li, Y., and Lyu, S. (2018). Exposing
deep fakes using inconsistent head poses. CoRR,
abs/1811.00661.
Yisroel Mirsky, W. L. (2020). The creation and detection of
deepfakes: A survey.
Zakharov, E., Shysheya, A., Burkov, E., and Lempitsky, V.
(2019). Few-shot adversarial learning of realistic neu-
ral talking head models.
SECRYPT 2021 - 18th International Conference on Security and Cryptography
784