Xinyuan Song

Generative AI for Medical Image Synthesis: GANs, Diffusion Models, and Data Augmentation Techniques

Yu Ding, Xudong Han, Junjie Yang, Tianyang Wang et autres

Generative models have emerged as a powerful paradigm for medical image synthesis and augmentation. The growing demand for fine-quality medical images has generated a significant amount of research attention, and these models are expected to address issues such as data sparsity, privacy …

ca, gb, cn, us, it, tw, ie, my (code pays fourni par la source)

CD9-positive tumor-associated macrophages promote renal cell carcinoma progression by activating the orphan nuclear receptor Nor1/Nr4a3

Pengyu Pan, Shiyuan Li, Wanru Yin, Peng Wang et autres

Tumor growth, metastasis, and treatment response in renal cell carcinoma (RCC) are believed to be regulated by the tumor microenvironment (TME). However, the mechanisms underlying the genomic, transcriptomic, and epigenetic alterations in clear cell RCC progression remain incompletely understood. This study analyzed …

cn (code pays fourni par la source)

Speculative Decoding in Decentralized LLM Inference: Turning Communication Latency into Computation Throughput

Jingwei Song, Wanyi Chen, Xinyuan Song, Max et autres

Speculative decoding accelerates large language model (LLM) inference by using a lightweight draft model to propose tokens that are later verified by a stronger target model. While effective in centralized systems, its behavior in decentralized settings, where network latency often dominates compute, …

Improvement mechanism of cellulose nanocrystal in coordination with calcium ions on the thermal gelation of soybean protein amyloid fibrils

Shanlong Zhu, Xinyuan Song, Kang Min Zhong, Lin Lü et autres

This study investigated the mechanism by which cellulose nanocrystals (CNC) derived from pear peel pomace synergize with Ca 2+ to regulate the thermal gelation of soy protein amyloid fibrils (SAFs). Results showed that the combined use of 0.09 % ( w / …

cn (code pays fourni par la source)

Sirp-α Antibody Inhibits Renal Cell Carcinoma Progression via Akt1/Akt2 Modulation in Tumor-Associated Macrophages

Junfeng Hao, Naiquan Liu, Xin Huang, Hanrong Li et autres

Introduction and Aim: Immunotherapies targeting tumor-associated macrophages (TAM) to improve antitumor immunity, are promising treatment strategies for many types of cancer. The signal-regulatory protein-α (Sirp-α)/CD47 axis is a key innate immune checkpoint target important in regulating phagocytosis in macrophages. We aimed to …

cn, us (code pays fourni par la source)

Lattica: A Decentralized Cross-NAT Communication Framework for Scalable AI Inference and Training

Jason Liu, Suli Wang, Xinyuan Song, Lei Ai et autres

The rapid expansion of distributed Artificial Intelligence (AI) workloads beyond centralized data centers creates a demand for new communication substrates. These substrates must operate reliably in heterogeneous and permissionless environments, where Network Address Translators (NATs) and firewalls impose significant constraints. Existing solutions, …

StructPrune: Structured Global Pruning asymptotics with $\mathcal{O}(\sqrt{N})$ GPU Memory

Xinyuan Song, Guangji Bai, Liang Zhao

Pruning is critical for scaling large language models (LLMs). Global pruning achieves strong performance but requires $\mathcal{O}(N)$ memory, which is infeasible for billion-parameter models. Local pruning reduces GPU memory usage to that of a single layer by pruning layers independently, but it …

Online Change-Point Detection for Functional Data

Hanbing Zhu, Houlin Zhou, Yehua Li, Xinyuan Song

In this article, we propose a CUSUM-type online change-point detection procedure for monitoring a change in the mean function of dependent functional data. Our method is fully nonparametric and does not require dimension reduction for the functional observations. For the proposed sequential …

cn, hk (code pays fourni par la source)

MountainLion: A Multi-Modal LLM-Based Agent System for Interpretable and Adaptive Financial Trading

Siyi Wu, Junqiao Wang, Xinyuan Song, Dapeng Yu et autres

Cryptocurrency trading is a challenging task requiring the integration of heterogeneous data from multiple modalities. Traditional deep learning and reinforcement learning approaches typically demand large training datasets and encode diverse inputs into numerical representations, often at the cost of interpretability. Recent progress …

ContrastiveGaussian: High-Fidelity 3D Generation with Contrastive Learning and Gaussian Splatting

Junbang Liu, Enpei Huang, Dongxing Mao, Hui Zhang et autres

Creating 3D content from single-view images is a challenging problem that has attracted considerable attention in recent years. Current approaches typically utilize score distillation sampling (SDS) from pre-trained 2D diffusion models to generate multi-view 3D representations. Although some methods have made notable …

hk, sg, us (code pays fourni par la source)

Performance analysis of large language models Chatgpt-4o, OpenAI O1, and OpenAI O3 mini in clinical treatment of pneumonia: a comparative study

Zhiwu Lin, Yuanyuan Li, Min Wu, Hongmei Liu et autres

This study aimed to compare the performance of three large language models (ChatGPT-4o, OpenAI O1, and OpenAI O3 mini) in delivering accurate and guideline compliant recommendations for pneumonia management. By assessing both general and guideline-focused questions, the investigation sought to elucidate each …

cn (code pays fourni par la source)

ContrastiveGaussian: High-Fidelity 3D Generation with Contrastive Learning and Gaussian Splatting

Junbang Liu, Enpei Huang, Dongxing Mao, Hui Zhang et autres

Creating 3D content from single-view images is a challenging problem that has attracted considerable attention in recent years. Current approaches typically utilize score distillation sampling (SDS) from pre-trained 2D diffusion models to generate multi-view 3D representations. Although some methods have made notable …