Partially-Connected Differentiable Architecture Search for Deepfake and Spoofing Detection

View on arXiv ← Back to list

Authors: Wanying Ge, Michele Panariello, Jose Patino, Massimiliano Todisco, Nicholas Evans

Published: 2021-04-07 13:53:20+00:00

AI Summary

This paper presents the first successful application of Partially-Connected Differentiable Architecture Search (PC-DARTS) to deepfake and spoofing detection. PC-DARTS efficiently learns complex neural architectures composed of convolutional operations and residual blocks, resulting in competitive performance with less computational complexity than existing state-of-the-art methods.

Abstract

This paper reports the first successful application of a differentiable architecture search (DARTS) approach to the deepfake and spoofing detection problems. An example of neural architecture search, DARTS operates upon a continuous, differentiable search space which enables both the architecture and parameters to be optimised via gradient descent. Solutions based on partially-connected DARTS use random channel masking in the search space to reduce GPU time and automatically learn and optimise complex neural architectures composed of convolutional operations and residual blocks. Despite being learned quickly with little human effort, the resulting networks are competitive with the best performing systems reported in the literature. Some are also far less complex, containing 85% fewer parameters than a Res2Net competitor.

Key findings

The automatically learned PC-DARTS models achieved competitive performance compared to state-of-the-art methods, with some models having 85% fewer parameters than a Res2Net competitor. The best performing model achieved a min-tDCF of 0.0914 and an EER of 4.96% on the ASVspoof 2019 LA evaluation partition.

Approach

The authors use PC-DARTS, a variant of differentiable architecture search, to automatically learn the optimal neural network architecture for deepfake detection. This approach uses random channel masking to reduce GPU time and optimizes both architecture and parameters via gradient descent. The learned architecture is then further optimized in a 'train from scratch' stage.

Datasets

ASVspoof 2019 Logical Access (LA) database

Model(s)

PC-DARTS (a neural architecture search method) resulting in networks composed of convolutional operations and residual blocks.

Author countries

France

← Previous