Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in Coakley et alK. L. Coakley, T. Snelleman, H. Hoos, and O. E. Gundersen, "The embrace of open science: An analysis of a decade of AI research and 56 800 conference papers," Under Review, 2026..

Backpropagation-Free Test-Time Adaptation via Probabilistic Gaussian Alignment

Authors: Youjia Zhang, Youngeun Kim, Young-Geun Choi, Hongyeob Kim, Huiling Liu, Sungeun Hong

NeurIPS 2025 | Venue PDF | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Extensive experiments across diverse benchmarks demonstrate that our method achieves state-of-the-art performance under a wide range of distribution shifts with superior scalability and robustness.
Researcher Affiliation Collaboration Youjia Zhang1 Youngeun Kim2 Young-Geun Choi1 Hongyeob Kim1 Huiling Liu1 Sungeun Hong1 1Sungkyunkwan University 2Amazon
Pseudocode Yes Algorithm 1 ADAPT: Online TTA... Algorithm 2 ADAPT: Transductive TTA
Open Source Code No The code will be fully publicly available upon acceptance of the paper.
Open Datasets Yes Dataset. We evaluated the proposed ADAPT on three different tasks: natural distribution shift, fine-grained categorization, and corruption robustness. Specifically, for natural distribution shift, we use multiple datasets including Image Net [4], Image Net-A [17], Image Net-R [15],Image Net-V [40], and Image Net-Sketch [50]. The corruption robustness task is evaluted on Image Net-C [16], which contains 15 corruption types covering noise, blur, weather, and digital artifacts. We also evaluate performance on 10 fine-grained recognition datasets: Aircraft [32], Caltech101 [8], Cars [22], DTD [3], Euro SAT [14], Flower102 [36], Food101 [1], Pets [37], SUN397 [53] and UCF101 [45].
Dataset Splits Yes The datasets we have used are all publicly available, including the data splits.
Hardware Specification Yes All experiments are conducted on an NVIDIA RTX A6000 GPU.
Software Dependencies No The paper does not explicitly list specific software dependencies with version numbers (e.g., Python, PyTorch, CUDA versions).
Experiment Setup Yes We set the coefficient α to 0.9, and assign the knowledge bank size L as 16 for online and 6 for transductive evaluation.