SCI: A Spectrum Concentrated Implicit Neural Compression for Biomedical Data

Authors: Runzhao Yang, Tingxiong Xiao, Yuxiao Cheng, Qianni Cao, Jinyuan Qu, Jinli Suo, Qionghai Dai

AAAI 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental The experiments show SCI s superior performance to state-of-the-art methods including commercial compressors, data-driven ones, and INR based counterparts on diverse biomedical data. Experimentally, we evaluated SCI s performance and validated its wide applicability on biological and medical data.
Researcher Affiliation Academia Runzhao Yang1, Tingxiong Xiao1, Yuxiao Cheng1, Qianni Cao2, Jinyuan Qu1, Jinli Suo1, Qionghai Dai1 1Department of Automation, Tsinghua University, Beijing 100084, China 2Department of Electrical Engineering, Tsinghua University, Beijing 100084, China
Pseudocode No No explicit pseudocode or algorithm blocks were found.
Open Source Code Yes The source code can be found at https: //github.com/Richeal Young/Implicit Neural Compression.git.
Open Datasets Yes For medical data we used Hi P-CT (Walsh et al. 2021), a public dataset containing data volumes of human organs from the tissue to cellular scales, which are of large diversity in both appearance (e.g., structure, contrast, brightness, and noise levels).
Dataset Splits No The paper does not explicitly provide training/test/validation dataset splits or cross-validation details for its own method.
Hardware Specification No No specific hardware details (like GPU/CPU models, memory, or processor types) used for running the experiments are mentioned in the paper.
Software Dependencies No The paper mentions software like OpenCV, FFmpeg, Pytorch, and Compress AI but does not provide specific version numbers for these or other critical software dependencies.
Experiment Setup Yes The network includes 7 layers, fr = 2.2, the hyper parameter of sinusoidal frequency w0 = 20, and is optimized by Adamax (Kingma and Ba 2014) with learning rate of 0.001. During partitioning, we set M = 1 and amax = 50 by default empirically.