Deep study track
Jaber Notes
A deeper, math-heavy companion to the viva modules: 16 topic notes with full derivations — foundations through generative models — plus an engineer roadmap and a model zoo. Mark each note complete as you work through it.
Notes
Math Foundations
Linear algebra, calculus, probability, statistics, information theory.
Learning Theory
ERM, bias-variance, regularization, cross-validation, VC dimension, PAC.
Linear Models
OLS, Ridge, Lasso, Elastic Net, logistic & softmax regression, GLMs.
Optimization
GD convergence, SGD, momentum, Adam/AdamW, LR schedules, Newton/L-BFGS.
Evaluation & Validation
Precision/recall/F1, ROC/PR-AUC, calibration, imbalance, CV strategies.
Feature Engineering
Scaling, encoding, missing data, outliers, selection, TF-IDF, leakage.
Classical Algorithms
k-NN, Naive Bayes, decision trees, SVM, k-means, GMM + EM.
Ensemble Methods
Bagging, random forests, AdaBoost, GBM, XGBoost, LightGBM, stacking.
Unsupervised Learning
Hierarchical/DBSCAN clustering, PCA, SVD, LDA, ICA, t-SNE, UMAP.
Practical ML
Workflow, HPO, pipelines, SHAP/LIME, calibration, drift, MLflow.
Neural Network Fundamentals
MLPs, activations, losses, full backprop derivation, init, gradients.
Training Deep Networks
BatchNorm/LayerNorm, dropout, AdamW, augmentation, schedules, AMP.
CNNs
Convolution math, receptive fields, ResNet, transfer learning, detection.
RNNs & Sequences
RNN, BPTT, LSTM/GRU gates, seq2seq, attention, beam search.
Transformers
Scaled attention, multi-head, positional encoding, BERT/GPT/T5, LoRA.
Generative Models
VAE (ELBO), GAN (WGAN, mode collapse), normalizing flows, diffusion.