2026-10-12T19:22:27Z https://www.scidb.cn/oai

10.57760/sciencedb.31085 2026-01-19T10:10:12Z

2026-01-19 Data and code for "Entropy Change in Sampling Predicts Downstream Performance in Neural Networks" doi:10.57760/sciencedb.31085 en The DOI contains both the data and codes for the work "Entropy Change in Sampling Predicts Downstream Performance in Neural Networks".Two tasks are related:vision task. In "vision data and results.zip" we submit the data used to test the accuracy as well as the data to perform Husimi entropy. Meanwhile the calculated results are also contained in this zip file. In the file "calculate_imagenet_delta_s.py" and "figure-imagenet.ipynb" we provide the code for calculate the Husimi entropy and the figure plot.NOTE: We add a new version ultilizing the multiscale Husimi-entropy for practical applications.MRI task. In "MRI raw data and results data.zip"we provide the MRI k-space data used to train the network as well as the calculated Husimi entropy data. In file "zjj-fMRI.ipynb" provides the code used to train the network and analyze the results. entropy auditing; Husimi spectrogram; subsampling; acquisition design; vision; MIMO; MRI Xiu-Cheng Wang Jun-Jie Zhang Nan Cheng Long-Gang Pang Tai-Jiao Du Deyu Meng PUBLIC https://mit-license.org dataset Science Data Bank