IOAI Guide

A Guide to the International Olympiad in Artificial Intelligence

SOTA – AI Community · follows the IOAI 2026 Syllabus

Abstract

This guide follows the official IOAI syllabus topic by topic, in the order the syllabus lists them. Each module explains an idea, shows how to use it in code, and links resources and past olympiad tasks to practise on. Modules are written by the community; the ones that aren't ready yet show exactly what the syllabus expects. Mark modules as you read them, and your progress is saved in this browser.

4 parts 60 modules 0 of 60 done

TTheory1 module

How it works. Contestants should understand core concepts and theoretical underpinnings—the “why” behind AI. This may involve studying textbooks, courses, and other resources to delve into the mechanics that power AI algorithms. Breadth should be prioritized over depth in covering all relevant topics.

PPractice35 modules

What it does, when to use it, and how to implement it. Contestants should develop practical skills necessary to implement AI methods in code. This includes knowing how to use library functions effectively, call the method on a particular data, and interpret outputs.

BBoth24 modules

Theory and practice. Certain topics require knowledge of both theoretical principles and practical application.

1Foundational Skills & Classical Machine Learning

0 of 25 done

1.1Programming Fundamentals

  1. 1.1.1 Python Basics (Loops, Functions, etc.) P
  2. 1.1.2 NumPy and Pandas for Data Handling P
  3. 1.1.3 Matplotlib and Seaborn for Visualization P
  4. 1.1.4 Scikit-learn for ML P
  5. 1.1.5 PyTorch Basics P
  6. 1.1.6 Tensor (Multi-dimensional Array) Manipulation P
  7. 1.1.7 Training Models on CPU and GPU P

1.2Supervised Learning

  1. 1.2.1 Linear Regression B
  2. 1.2.2 Logistic Regression B
  3. 1.2.3 L1 & L2 Regularization B
  4. 1.2.4 K-Nearest Neighbors (K-NN) B
  5. 1.2.5 Decision Trees B
  6. 1.2.6 Model Ensembles (Gradient Boosting, Bagging, Random Forest) P
  7. 1.2.7 Support Vector Machines (SVM) B

1.3Unsupervised Learning

  1. 1.3.1 K-Means Clustering B
  2. 1.3.2 Principal Component Analysis (PCA) B
  3. 1.3.3 t-SNE, UMAP P
  4. 1.3.4 DBSCAN, Hierarchical & Spectral Clustering P

1.4Data Science Fundamentals

  1. 1.4.1 Model Evaluation Metrics (Accuracy, Precision, Recall, F1-Score, etc.) B
  2. 1.4.2 Underfitting, Overfitting T
  3. 1.4.3 Hyperparameter Tuning P
  4. 1.4.4 Cross-Validation P
  5. 1.4.5 Confusion Matrix and ROC Curves B
  6. 1.4.6 Feature Engineering P
  7. 1.4.7 Data Processing P

2Neural Networks & Deep Learning

0 of 18 done

2.1Neural Networks

  1. 2.1.1 Perceptron Basics B
  2. 2.1.2 Gradient Descent B
  3. 2.1.3 Backpropagation B
  4. 2.1.4 Activation Functions (ReLU, Sigmoid, Tanh) B
  5. 2.1.5 Loss Functions (MSE, MAE, Cross Entropy, etc.) B

2.2Deep Learning

  1. 2.2.1 Multi-Layer Perceptrons (MLP) B
  2. 2.2.2 Data Embeddings (text, image, audio) B
  3. 2.2.3 Pooling Techniques (Max, Average) B
  4. 2.2.4 Attention Mechanism B
  5. 2.2.5 Transformers (theory needed only for text and image) B
  6. 2.2.6 Autoencoders P
  7. 2.2.7 SGD, Mini-Batch Gradient Descent B
  8. 2.2.8 Momentum Methods (Adam, AdamW) P
  9. 2.2.9 Convergence and Learning Rates P
  10. 2.2.10 Regularization: Dropout, Early Stopping, Weight Decay P
  11. 2.2.11 Weight Initialization P
  12. 2.2.12 Batch Normalization P
  13. 2.2.13 Model Finetuning (full and parameter-efficient) P

3Computer Vision

0 of 10 done

3.1Fundamentals

  1. 3.1.1 Convolutional Layers B
  2. 3.1.2 Image Classification P
  3. 3.1.3 Object Detection (YOLO, SSD, DETR) P
  4. 3.1.4 Image Segmentation (U-Net) P
  5. 3.1.5 Pre-trained Vision Encoders (e.g. ResNet) P
  6. 3.1.6 Image Augmentation P
  7. 3.1.7 Generating Images with GANs P
  8. 3.1.8 Self-Supervised Learning for Vision P
  9. 3.1.9 Vision-text encoders (e.g. CLIP) P
  10. 3.1.10 Diffusion Models P

4Natural Language Processing & Audio

0 of 7 done

The data can be text, tabular, image, audio, video, and time-series, and should be processed with the methods above.

4.1NLP

  1. 4.1.1 Text Classification P
  2. 4.1.2 Pre-trained Text Encoders (e.g. BERT) B
  3. 4.1.3 Language Modeling B
  4. 4.1.4 Encoder-Decoder Models (e.g. for Machine Translation or Vision-Language Modeling) P
  5. 4.1.5 Pre-trained Language Models (open-source and API-based ones) P

4.2Audio Processing

  1. 4.2.1 Pre-trained Audio Encoders: HuBERT P
  2. 4.2.2 Audio Models: Qwen-Audio, Whisper, Voxtral P