Currently, pretrained models are rapidly scaling in size, which substantially increases the cost of fine-tuning them for downstream tasks. To address this challenge, parameter-efficient fine-tuning (PEFT) methods have been developed to optimize a minimal set of parameters for adaptation. While current PEFT approaches predominantly employ an "additive'' strategy, introducing learnable modules into inputs or architectures, neglect the inherent knowledge embedded within pretrained models, which may be redundant or even conflict with downstream tasks. This limitation leads to increased inference latency and suboptimal transfer performance, particularly in scenarios with significant domain gaps. In this paper, we propose a Subtractive Fine-tuning Paradigm(SFP), which converts multiple redundant operations within the original module into a linear transformation to enhance inference speed and model performance. Specifically, we introduce a compact filter block to replace specific module with interference and redundancy in the original structure to reduce model conflicts. By using a pseudo inverse matrix to construct filter block, ensuring that it can inherit the knowledge of the replacement module, and then freezing the rest of the model, only fine-tuning the filter block is performed to eliminate interference and redundant knowledge, thereby enhancing the model’s adaptability to downstream tasks. Experimental results demonstrate that our SFP outperforms existing PEFT methods in accuracy while decreasing the overall model parameters by 12%. Compared to full fine-tuning, the accuracy has increased by 8.47%(74.04% vs. 65.57%, VTAB).

The significance of seismic diffractions for the high-resolution imaging of subsurface discontinuities has been emphasized in recent years. Separating diffractions from strong specular reflected wavefields is a crucial process owing to the weak amplitude of diffractions. The low-rank (LR) characteristics of seismic data have been successfully implemented for diffracted wavefield isolation using rank-reduction methods. Traditional LR-based diffraction separation uses the optimal LR approximation of the Hankel matrix formulated from seismic data for predicting linear reflection events. However, without the Hankel structure in traditional LR approximation, the Hankel matrix of estimated reflections does not exhibit the expected LR properties, which may affect the predicted reflection accuracy. In this study, a regularized LR (RLR) approximation method that exploits the LR properties of reflection events and the corresponding Hankel structure was developed to enhance diffractions and eliminate reflections. The RLR approximation algorithm considered the LR constraint of the Hankel matrix for estimated reflections, leading to improved LR approximation. Synthetic and field examples were used to demonstrate the effectiveness of the proposed algorithm in separating diffractions and imaging small subsurface geological structures.