Search papers, labs, and topics across Lattice.
This paper introduces a curvature-aware zeroth-order (CAZO) optimization method for test-time adaptation (TTA) that enhances the performance of pre-trained models on unlabeled test data without the need for backpropagation. By leveraging the low-rank Hessian structure of the loss landscape, CAZO employs a covariance matrix for anisotropic perturbation sampling, allowing for more accurate gradient estimation while freezing pretrained weights and optimizing minimal adapter parameters. The proposed method achieves state-of-the-art performance in TTA, significantly reducing memory overhead compared to traditional backpropagation-based approaches.
CAZO achieves state-of-the-art test-time adaptation performance while slashing memory requirements, making it a game-changer for on-device AI applications.
Test-time adaptation (TTA) aims to enhance the cross-domain performance of pre-trained models by adapting to unlabeled test data. While most existing TTA methods rely on backpropagation (BP) for finetuning, BP-free methods such as zeroth-order (ZO) methods are more desired in practical on-device scenarios. ZO methods rely only on forward computation, which can largely reduce the complexity and memory overhead of on-device deployment. However, ZO methods suffer from much higher variance compared with first-order methods in estimating the gradient. To address this, we propose an improved ZO method to substantially boost the performance of ZO optimization based TTA. First, we provide an observation to reveal the persistent low-rank Hessian structure of the loss during the adaptation process. Based on this insight, we then propose a loss-landscape curvature-aware zeroth-order (CAZO) method, which leverages a sliding-average estimation of the diagonal Hessian to construct a covariance matrix for anisotropic perturbation sampling. CAZO operates by freezing pretrained weights and optimizing minimal adapter parameters via forward-only passes based gradient estimation, which can substantially reduce the memory overhead compared to BP-based methods. Extensive experiments demonstrate that CAZO significantly outperforms existing TTA methods, achieving state-of-the-art performance while maintaining an excellent balance between accuracy and memory efficiency. Code is available at https://github.com/Hollyming/CAZO.
