Akademik Dersler - İçerik

Can explain the fundamental theoretical concepts of machine learning and optimization algorithms with their mathematical foundations.
Can distinguish between supervised and unsupervised learning paradigms and match them with appropriate problem types.
Can describe the structure of loss/cost functions and the calculus foundations behind Gradient Descent optimization.
Can implement fundamental machine learning algorithms from scratch without using ready-made libraries.
Can encode vector and matrix operations in a programming language using vectorization.
Can translate the forward propagation and backpropagation steps of basic architectures into code using only linear algebra equations.
Can apply machine learning models using industry-standard tools.
Can train classical machine learning algorithms such as Logistic Regression, Support Vector Machines, and Decision Trees on datasets.
Can perform clustering analysis on unlabeled data using the distance-based K-Means algorithm.
Can evaluate model quality using scientific metrics and apply data preprocessing techniques.
Can reduce high-dimensional data using eigenvalue/eigenvector decomposition through Principal Component Analysis.
Can detect overfitting and the bias–variance tradeoff using K-fold cross-validation and perform hyperparameter optimization.
Can design and train deep learning architectures using modern frameworks.
Can construct dynamic computation graphs using tensors and utilize automatic differentiation mechanisms.
Can build data preparation pipelines and basic training loops for Convolutional Neural Networks in computer vision problems.