luweiagi/machine-learning-notes

What it is. A collection of study notes on machine learning and related topics. The README is mostly a huge table of contents: it lists dozens of note pages grouped under five big sections (Preface, Math Foundations, Programming Foundations, Machine Learning, Deep Learning), with each entry linking to a separate note file. The project also points to an external "machine learning algorithm map" (a diagram on another site) and includes a PNG image of that map.

How it's meant to be read. The README begins with a warning (in Chinese): if you're viewing the page on github.com, stop reading there and click the link at the top, because you're in source mode and the math formulas aren't rendered. The intended reading experience is via the GitHub Pages site at https://luweiagi.github.io/machine-learning-notes/, where the formulas display properly.

What's covered (from the table of contents).

• Preface — personal intro, ML intro, a suggested learning path, interview tips, and a history of machine learning.
• Math foundations — calculus (Taylor expansion, e, Fourier transform, Hilbert space, convolution), linear algebra (norms, matrix derivatives, eigenvalues, SVD), probability & statistics (variance/covariance, Bayesian analysis, probability distributions like Normal, Gamma, Beta, Dirichlet), complex functions, information theory, numerical optimization (least squares, Lagrange multipliers, convex optimization), and gradient-update algorithms (SGD, Momentum, Nesterov, AdaGrad, RMSprop, Adadelta, Adam, Nadam, AMSGrad, and more).
• Programming foundations — computer science course topics (computer organization, data structures & algorithms, compilers, operating systems, computer networks), programming languages (C/C++ with STL, multithreading; Python with syntax, libraries like numpy/pandas/scipy/matplotlib/scikit-learn, Cython), big data tools (Hadoop, Hive, Spark), and data backup tools (rsync, Git).
• Machine learning — overview and philosophy, feature engineering, perceptron, k-nearest neighbor, naive Bayes, linear models (linear/logistic regression, max entropy), decision trees, support vector machines (including SMO), liblinear, ensemble methods (Bagging, random forests, Boosting, AdaBoost, GBDT, XGBoost, LightGBM, Stacking), dimensionality reduction (PCA, manifold learning; LDA/t-SNE/ICA/autoencoders appear as plain list items), EM algorithm, probabilistic graphical models, HMM, CRF, clustering (k-means, Gaussian mixtures, DPMM), topic models (LDA), and assorted tips (loss functions, negative sampling).
• Deep learning — overview, artificial neural networks, Hopfield networks, and tips (feature crossing, activation functions, normalization like Batch/Layer Norm, backpropagation, gradient explosion/vanishing, gradient clipping).

Organization. The notes are nested in a hierarchical folder structure that mirrors the table of contents — e.g., mathematics/calculus/taylor-expansion/taylor-expansion.md, machine-learning/support-vector-machine/smo/smo.md. The README itself is just an index; the actual content lives in those linked .md files.

Notable details.

• The list contains a few topics that appear as plain text without hyperlinks (e.g., AdamW, LDA linear discriminant analysis, t-SNE, ICA, self-encoders) — the README lists these as items but does not link them to a note file.
• The project is written primarily in Chinese.
• The README contains no description of the author's intent beyond the structure of the notes themselves — no stated goals, license, or contribution guidelines appear.