Thelossesarecompared onthreedegreesofshift(easy,moderate and hard), which is controlled by the drifted distance of Gaussian clusters. Herewediscuss the architecture chosen and the implementation details. Note that the task loss / surrogate loss function is used to update the meta-loss mϕ during meta-learning. The number of transformer layers and the hidden layers in MLP are selected from{1,2}. Wecanseethatthetask loss barely affects the learnt meta loss.

Adapting to dynamic data distributions is a practical yet challenging task.

Convolutional Neural Networks (CNNs), as the predominant architecture in deep learning, play a crucial role in image, video, and audio processing [1, 2, 3]. CNNs were originally inspired by the concept of receptive fields in the biological visual system [4], and our focus is to explore and leverage similar characteristics within CNNs.

Current test-time adaptation (TT A) approaches aim to adapt a machine learning model to environments that change continuously.

In recent years, deep neural networks (DNNs) have witnessed extensive applications, and protecting their intellectual property (IP) is thus crucial. As a noninvasive way for model IP protection, model fingerprinting has become popular.