RTF-Q: Efficient Unsupervised Domain Adaptation with Retraining-free Quantization

N Du, C Tang, Y Jiang, Y Meng, Z Wang - arxiv preprint arxiv:2408.05752, 2024 - arxiv.org
Performing unsupervised domain adaptation on resource-constrained edge devices is
challenging. Existing research typically adopts architecture optimization (eg, designing …

[PDF][PDF] A Subdomain-Specific Knowledge Distillation for Unsupervised Domain Adaptation in Adverse Weather Conditions

G Choi, Y Lee, D Jang, DS Kim - International Conference on …, 2023 - valeoai.github.io
Method. The architecture consists of two subdomain teachers, a student, and a student
exponential moving average (EMA) model. Our method performs self-training on each …

Resource Efficient Test-Time Training with Slimmable Network

L Cai, S Li, W Ma, L Liu - openreview.net
Test-Time Training (TTT), an innovative paradigm for enhancing a model's generalization in
a specific future scenario, commonly leverages self-supervised learning to adapt the model …