YCYuxin Cheng

Yuxin Cheng

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Yuxin Cheng has a total of 16 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Hardware-aware Low-Rank Adaptation for Large Language Models Based on Hybrid Compute-in-Memory Architecture

    by Taiqiang Wu, Chenchen Ding, Wenyong Zhou, Yuxin Cheng, Xincheng Feng, Shuqi Wang, Wendong Xu, Chufan Shi, Zhengwu Liu, Ngai Wong

    Low-rank adaptation (LoRA) is a predominant parameter-efficient finetuning method for adapting large language models (LLMs) to downstream tasks. Meanwhile, Compute-in-Memory (CIM) architectures demonstrate superior energy efficiency due to their array-level parallel in-memory computing designs. In this article, we propose deploying the LoRA-finetuned LLMs on the hybrid CIM architecture (i.e., pretrained weights onto energy-efficient Resistive Random-Access Memory (RRAM) and LoRA branches onto noise-free Static Random-Access Memory (SRAM)), reducing the energy cost to about 3% compared with the Nvidia A100 GPU. However, the inherent noise of RRAM on the saved weights leads to performance degradation, simultaneously. To address this issue, we design a novel Hardware-aware Low-rank Adaptation (HaLoRA) method. The key insight is to train a LoRA branch that is robust toward such noise and then deploy it on noise-free SRAM, while the extra cost is negligible since the parameters of LoRAs are much fewer than pretrained weights (e.g., 0.15% for LLaMA-3.2 1B model). To improve the robustness towards the noise, we theoretically analyze the gap between the optimization trajectories of the LoRA branch under both ideal and noisy conditions and further design an extra loss to minimize the upper bound of this gap. Therefore, we can enjoy both energy efficiency and accuracy during inference. Experiments finetuning the Qwen and LLaMA series demonstrate the effectiveness of HaLoRA across multiple reasoning tasks, achieving up to 22.7 improvement in average score while maintaining robustness at various noise types and noise levels.


  • Open Access

    Re-Activating Frozen Primitives for 3D Gaussian Splatting

    by Yuxin Cheng, Binxiao Huang, Wenyong Zhou, Taiqiang Wu, Zhengwu Liu, Graziano Chesi, Ngai Wong


  • Open Access

    Instance by Instance: An Iterative Framework for Multi-Instance 3D Registration

    by Jiaqi Yang, Xinyue Cao, Xiyu Zhang, Yuxin Cheng, Zhaoshuai Qi, Siwen Quan

    Multi-instance registration is a challenging problem in computer vision and robotics, where multiple instances of an object need to be registered in a standard coordinate system. Pioneers followed a non-extensible one-shot framework, which prioritizes the registration of simple and isolated instances, often struggling to accurately register challenging or occluded instances. To address these challenges, we propose the first iterative framework for multi-instance 3D registration (MI-3DReg) in this work, termed instance-by-instance (IBI). It successively registers instances while systematically reducing outliers, starting from the easiest and progressing to more challenging ones. This enhances the likelihood of effectively registering instances that may have been initially overlooked, allowing for successful registration in subsequent iterations. Under the IBI framework, we further propose a sparse-to-dense correspondence-based multi-instance registration method (IBI-S2DC) to enhance the robustness of MI-3DReg. Experiments on both synthetic and real datasets have demonstrated the effectiveness of IBI and suggested the new state-of-the-art performance with IBI-S2DC, e.g., our mean registration F1 score is 12.02%/12.35% higher than the existing state-of-the-art on the synthetic/real datasets. The source codes are availableonline at https://github.com/caoxy01/IBI.