All Items

• Compression-Based Tokenization Improves Language Modeling of Hierarchical Genomic Structure

  Authors: Yiping Wang, Jing Wang, Junhao Zhu, Fengyao Zhai, Hu Zhu, Ziwei Dai, Zengru Di, Da Zhou, Yu Liu

  Abstract:

  Tokenization is a critical design choice in genomic language modeling. Widely used schemes---character-level encoding, fixed-length $k$-mers, and greedy subword algorithms such as BPE---show intrinsic limitations on DNA that are magnified by the small four-letter alphabet. To address this, we adapt Ladderpath, an Algorithmic Information Theory method that identifies nested and hierarchical...

  Keywords: Tokenization; Algorithmic Information Theory; Language Model

  Timeline: First Posted 2025-12-11 | Last Updated 2025-12-11

  Metrics: Views: 193  |  Downloads: 89  |  Favorites: 0

• Performance Evaluation of the SmartBrain: A Wearable PET System for Human Brain Imaging

  Authors: Han Liu, Wenkang Qu, Da Liang, Xin Yu, Yuejie Lin, Haoyu Zou, Siyuan Han, Zhijun Zhao, Ying Lin, Xiaoyin Zhang, Jinyong Tao, Wenbin Li, Huiping Zhao, Yibin Zhang, Gongning Luo, Ningyi Jiang, Qiyu Peng

  Abstract:

  Background. Positron Emission Tomography (PET) is a vital medical imaging tool for studying in vivo metabolism. However, conventional PET systems are limited in their inability to image during free movement. Therefore, we have developed a wearable brain PET system for real-time imaging, called SmartBrain. Methods. The...

  Keywords: Wearable PET; Human brain imaging; NEMA; SmartBrain

  Timeline: First Posted 2025-12-04 | Last Updated 2025-12-04

  Metrics: Views: 280  |  Downloads: 131  |  Favorites: 0