Hacettepe University · Fall 2026
A broad, technically grounded introduction to the ideas and algorithms behind modern machine learning, from nearest neighbors and linear models to neural networks, kernels, ensembles, clustering, and dimensionality reduction.
Course information
AID202 is an undergraduate introductory course in machine learning. It gives a broad overview of core concepts and algorithms, ranging from supervised learning methods such as support vector machines and decision trees to unsupervised learning methods such as clustering and dimensionality reduction. The goal is to develop both a solid understanding of the underlying ideas and the ability to apply them to real-world problems.
The course assumes comfort with programming, calculus, linear algebra, and basic probability and statistics. We will use these foundations to reason about model fitting, generalization, optimization, representation, and evaluation.
References
No single book is required. The schedule points to readings from several references when they are useful.
Policies
Unless stated otherwise, all work on assignments must be completed individually. You are encouraged to discuss concepts with classmates, but discussions about a specific solution, including code or pseudocode, are not allowed. Submitting someone else’s work, in whole or in part, as your own is a violation of academic integrity. The same principle applies to material found online.
The permitted use of generative AI tools may differ across assignments and course activities. The specific policy will be stated for each assignment. Students remain responsible for understanding, verifying, and being able to explain all work they submit. Generative AI tools may not be used during examinations unless explicitly stated otherwise.
Assessment
Fall 2026
What is machine learning? Problem types, examples, similarity, and the k-nearest neighbor classifier.
Reading: The Discipline of Machine Learning, Tom Mitchell; Barber 1, 14.1–14.2.
Videos: The Master Algorithm, Pedro Domingos; The Thinking Machine.
Demo: k-Nearest Neighbors.
From local prediction to parametric models, then generalization, model complexity, and regularization.
Reading: Bishop 1.1, 3.1; Stanford CS229 notes.
Demo: Curve fitting.
Evaluation, overfitting, train/validation/test practice, basic learning theory, and probability foundations.
Reading: P. Domingos, A few useful things to know about machine learning; Daumé III Ch. 12; Barber 1.1–1.4.
Background: E. Simoncelli, A Geometric Review of Linear Algebra.
Demo: Seeing Theory.
Likelihood-based estimation, priors, posterior reasoning, and generative classification.
Reading: Murphy 2.1–2.3.2; Daumé III Ch. 7; Tom Mitchell, Naïve Bayes.
Optional: Learning to Decode Cognitive States from Brain Images.
Demo: Bayes theorem.
Linear decision boundaries and two complementary views of probabilistic classification.
Reading: SLP3 Ch. 5; Bishop 4.1.1–4.1.2, 4.5; Daumé III Ch. 3.
Optional: On Discriminative vs. Generative Classifiers, Ng and Jordan.
Each project group will briefly present its proposed project to the class. We will discuss the motivation, problem formulation, proposed methodology, datasets, evaluation strategy, and scope. The goal is to use feedback from the whole class to refine and better position each project before the main implementation phase begins. All students are expected to participate actively in the discussion and provide constructive feedback to other groups.
From linear units to layered networks, then efficient gradient computation.
Reading: Bishop 5.1; CS231n backpropagation notes.
Demo: A Neural Network Playground.
Representation learning, depth, convolution, and the inductive biases behind CNNs.
Reading: LeCun, Bengio, and Hinton, Deep Learning.
Christopher Olah: Conv Nets: A Modular Perspective and Understanding Convolutions.
The exact exam timing and in-class format will be confirmed during the semester.
Maximum-margin classification, slack variables, and extending linear SVMs beyond the binary separable setting.
Reading: Alpaydin 13.1–13.3, 13.9; M. A. Hearst, Support Vector Machines; CS229 notes.
Video: Patrick Winston, Support Vector Machines.
Demo: Multi-class SVM.
Nonlinear decision functions through kernels, then recursive partitioning and tree-based prediction.
Reading: Alpaydin 13.5–13.7, 13.10; Mitchell Ch. 3; Bishop 14.4.
Combining learners to reduce variance or bias, with tree ensembles and AdaBoost.
Reading: Bishop 14.1–14.3; Understanding the Bias-Variance Tradeoff.
Optional: Real-Time Human Pose Recognition in Parts from Single Depth Images; Rapid Object Detection using a Boosted Cascade of Simple Features.
Video: A Boosting Tutorial, Robert Schapire.
K-means, spectral clustering, agglomerative clustering, PCA, SVD, ICA, and autoencoders.
Reading: Bishop 9.1; Cluster Analysis: Basic Concepts and Algorithms, Tan, Steinbach, and Kumar.
Student project presentations, synthesis of the course, and connections across the methods covered during the semester.
Reference shelf