Li Xiong

Manlai Liang, JiaMing Zhang, Xiong Li et al.

The increasing size of the Key-Value (KV) cache during the Large Language Models long-context inference is the main obstacle for its balance between the deployment cost and task accuracy. To reduce the KV cache size in such scenarios, most previous efforts leveraged on the attention weight to evict non-critical cache tokens. But there is a trade-off in those methods, they usually require major modification of the inference infrastructure and significant computation overhead. Based on the fact that the Large Language models are autoregressive models, we propose LagKV, a KV compression strategy only relying on straight forward comparison among KV themselves. It is a totally attention free method which offers easy integration to the main stream inference platform and comparable performance comparing to other complicated KV compression methods. Results on RULER benchmark show that, our approach outperforms SnapKV and StreamingLLM in different compression ratios. Especially in the 64-digit passkey retrieval task, our method outperforms the attention weight based method $H_2O$ over $50\%$ with same compression ratios. Our code is available at https://github.com/AI-Lab-China-Merchants-Bank/LagKV.

Yuanyuan Peng, Pengpeng Luan, Hongbin Tu et al.

Priori knowledge of pulmonary anatomy plays a vital role in diagnosis of lung diseases. In CT images, pulmonary fissure segmentation is a formidable mission due to various of factors. To address the challenge, an useful approach based on ODoS filter and shape features is presented for pulmonary fissure segmentation. Here, we adopt an ODoS filter by merging the orientation information and magnitude information to highlight structure features for fissure enhancement, which can effectively distinguish between pulmonary fissures and clutters. Motivated by the fact that pulmonary fissures appear as linear structures in 2D space and planar structures in 3D space in orientation field, an orientation curvature criterion and an orientation partition scheme are fused to separate fissure patches and other structures in different orientation partition, which can suppress parts of clutters. Considering the shape difference between pulmonary fissures and tubular structures in magnitude field, a shape measure approach and a 3D skeletonization model are combined to segment pulmonary fissures for clutters removal. When applying our scheme to 55 chest CT scans which acquired from a publicly available LOLA11 datasets, the median F1-score, False Discovery Rate (FDR), and False Negative Rate (FNR) respectively are 0.896, 0.109, and 0.100, which indicates that the presented method has a satisfactory pulmonary fissure segmentation performance.

Yuanyuan Peng, Zixu Zhang, Hongbin Tu et al.

Background: The 2019 novel coronavirus disease (COVID-19) has been spread widely in the world, causing a huge threat to people's living environment. Objective: Under computed tomography (CT) imaging, the structure features of COVID-19 lesions are complicated and varied greatly in different cases. To accurately locate COVID-19 lesions and assist doctors to make the best diagnosis and treatment plan, a deep-supervised ensemble learning network is presented for COVID-19 lesion segmentation in CT images. Methods: Considering the fact that a large number of COVID-19 CT images and the corresponding lesion annotations are difficult to obtained, a transfer learning strategy is employed to make up for the shortcoming and alleviate the overfitting problem. Based on the reality that traditional single deep learning framework is difficult to extract COVID-19 lesion features effectively, which may cause some lesions to be undetected. To overcome the problem, a deep-supervised ensemble learning network is presented to combine with local and global features for COVID-19 lesion segmentation. Results: The performance of the proposed method was validated in experiments with a publicly available dataset. Compared with manual annotations, the proposed method acquired a high intersection over union (IoU) of 0.7279. Conclusion: A deep-supervised ensemble learning network was presented for coronavirus pneumonia lesion segmentation in CT images. The effectiveness of the proposed method was verified by visual inspection and quantitative evaluation. Experimental results shown that the proposed mehtod has a perfect performance in COVID-19 lesion segmentation.