Diversity-aware Weight Perturbation Promotes Robust Adaptation

Zibo Chen, Ruxin Li, Zilu Wang
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:15211-15227, 2026.

Abstract

Compute-In-Memory (CIM) accelerators are promising for energy-efficient edge inference, yet they faces fundamental challenges when deploying Deep Neural Networks (DNNs), as hardware-induced weight perturbations from intrinsic noise and device drift degrade accuracy and impede reliable inference. To tackle this challenge, we propose Diversity-aware Weight Perturbation (DWP), an immune-system-inspired training method that emulates affinity-based selection by exploiting sample-level prediction disagreement under diverse noise realizations to guide adaptive sample weighting, building robustness to weight perturbation. Experiments show that DWP-trained models consistently yield superior robustness, achieving over 15% accuracy improvements compared to standard-trained models under severe weight perturbations (mismatch level up to 70%) and maintaining inference accuracy at 90% over a simulated one-year CIM operation with only 2%–4% variation in accuracy. Moreover, under matched model and inference configurations, deployment on low-precision CIM hardware reduces inference energy by 38% compared to a GPU baseline. These results demonstrate that DWP enables robust and energy-efficient neural network deployment on resource-constrained edge devices with inherent hardware uncertainties.

Cite this Paper

BibTeX
@InProceedings{pmlr-v306-chen26bq, title = {Diversity-aware Weight Perturbation Promotes Robust Adaptation}, author = {Chen, Zibo and Li, Ruxin and Wang, Zilu}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {15211--15227}, year = {2026}, editor = {Zhang, Tong and Dudik, Miroslav and Jaggi, Martin and Agarwal, Alekh and Li, Sharon and Schuurmans, Dale and Zhu, Jerry and Berkenkamp, Felix and Dong, Hanze and Bietti, Alberto}, volume = {306}, series = {Proceedings of Machine Learning Research}, month = {06--11 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v306/main/assets/chen26bq/chen26bq.pdf}, url = {https://proceedings.mlr.press/v306/chen26bq.html}, abstract = {Compute-In-Memory (CIM) accelerators are promising for energy-efficient edge inference, yet they faces fundamental challenges when deploying Deep Neural Networks (DNNs), as hardware-induced weight perturbations from intrinsic noise and device drift degrade accuracy and impede reliable inference. To tackle this challenge, we propose Diversity-aware Weight Perturbation (DWP), an immune-system-inspired training method that emulates affinity-based selection by exploiting sample-level prediction disagreement under diverse noise realizations to guide adaptive sample weighting, building robustness to weight perturbation. Experiments show that DWP-trained models consistently yield superior robustness, achieving over 15% accuracy improvements compared to standard-trained models under severe weight perturbations (mismatch level up to 70%) and maintaining inference accuracy at 90% over a simulated one-year CIM operation with only 2%–4% variation in accuracy. Moreover, under matched model and inference configurations, deployment on low-precision CIM hardware reduces inference energy by 38% compared to a GPU baseline. These results demonstrate that DWP enables robust and energy-efficient neural network deployment on resource-constrained edge devices with inherent hardware uncertainties.} }
Endnote
%0 Conference Paper %T Diversity-aware Weight Perturbation Promotes Robust Adaptation %A Zibo Chen %A Ruxin Li %A Zilu Wang %B Proceedings of the 43rd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2026 %E Tong Zhang %E Miroslav Dudik %E Martin Jaggi %E Alekh Agarwal %E Sharon Li %E Dale Schuurmans %E Jerry Zhu %E Felix Berkenkamp %E Hanze Dong %E Alberto Bietti %F pmlr-v306-chen26bq %I PMLR %P 15211--15227 %U https://proceedings.mlr.press/v306/chen26bq.html %V 306 %X Compute-In-Memory (CIM) accelerators are promising for energy-efficient edge inference, yet they faces fundamental challenges when deploying Deep Neural Networks (DNNs), as hardware-induced weight perturbations from intrinsic noise and device drift degrade accuracy and impede reliable inference. To tackle this challenge, we propose Diversity-aware Weight Perturbation (DWP), an immune-system-inspired training method that emulates affinity-based selection by exploiting sample-level prediction disagreement under diverse noise realizations to guide adaptive sample weighting, building robustness to weight perturbation. Experiments show that DWP-trained models consistently yield superior robustness, achieving over 15% accuracy improvements compared to standard-trained models under severe weight perturbations (mismatch level up to 70%) and maintaining inference accuracy at 90% over a simulated one-year CIM operation with only 2%–4% variation in accuracy. Moreover, under matched model and inference configurations, deployment on low-precision CIM hardware reduces inference energy by 38% compared to a GPU baseline. These results demonstrate that DWP enables robust and energy-efficient neural network deployment on resource-constrained edge devices with inherent hardware uncertainties.
APA
Chen, Z., Li, R. & Wang, Z.. (2026). Diversity-aware Weight Perturbation Promotes Robust Adaptation. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:15211-15227 Available from https://proceedings.mlr.press/v306/chen26bq.html.

Related Material