Appendix E - Recommended Resources#
Books#
Zhou Zhihua, Machine Learning (in Chinese) - especially Chapter 2 on model evaluation and selection;
Hang Li, Statistical Learning Methods (in Chinese) - classic derivations of logistic regression and perceptrons;
Gonzalez & Woods, Digital Image Processing - bit planes, histograms, and frequency-domain fundamentals (early experiments used its images; v1.4 data sources switched to campus photos and BOSSbase);
Any Chinese textbook on information hiding and digital watermarking as broad reading.
Core Papers (in reading order)#
Westfeld, F5 - A Steganographic Algorithm, 2001 - matrix embedding and F5;
Westfeld & Pfitzmann, Attacks on Steganographic Systems, 1999 - chi-square test;
Fridrich, Goljan & Du, Reliable Detection of LSB Steganography in Color and Grayscale Images, 2001 - RS analysis;
Fridrich, Goljan, Lisonek & Soukal, Writing on Wet Paper, 2005 - wet paper coding;
Pevny, Filler & Bas, Using High-Dimensional Image Statistics to Perform Unsigned Steganalysis - modern feature-based steganalysis (SPAM) entry point.
Online#
The project README and the canonical results table
docs/RESULTS.md(current numbers for the SRM / 143-D / 53-D experiments); GitHub: Yukinoshita-lin/nsf5-steganography;scikit-learn docs: LogisticRegression, GroupKFold, roc_curve;
PyTorch tutorials (needed for the GPU pipeline);
GitHub Actions documentation for CI/release.
Try it | The handbook ends here. Return to the checklist in 0.4 and grade yourself item by item. If you can tick all of them, give yourself a certificate: nsF5 Steganography - From Zero to Understanding.