Academic Presentations

Comprehensive Data Science & Machine Learning Course Materials

Diogo Ribeiro
ESMAD - Escola Superior de Média Arte e Design
Lead Data Scientist, Mysense.ai

🎯 Overview

This repository contains a comprehensive collection of professional academic presentations covering advanced topics in statistics, machine learning, deep learning, and data science. The materials are designed for:

• 🎓 Graduate-level courses in Data Science, Statistics, and Computer Science
• 🔬 Research seminars and academic conferences
• 🏢 Professional training programs in industry
• 📚 Self-study for advanced learners

Key Features

✅ 15+ comprehensive presentations with 100+ hours of content
✅ Production-ready code in Python and R (27,000+ lines)
✅ 140+ curated references with DOIs
✅ Professional LaTeX theme with consistent styling
✅ Hands-on exercises and assessments
✅ Automated PDF generation via GitHub Actions

📚 Course Catalog & Learning Objectives

🔷 Deep Learning & Neural Networks

Deep Learning Fundamentals

📂 02-deep-learning/deep-learning-fundamentals/

Learning Objectives:

• Understand the mathematical foundations of neural networks
• Implement backpropagation and gradient descent from scratch
• Master modern optimization techniques (SGD, Adam, AdamW)
• Design and train CNN architectures for computer vision
• Build RNN/LSTM models for sequential data
• Understand Transformer architecture and attention mechanisms
• Apply regularization techniques (dropout, batch normalization)

Topics Covered:

• Perceptron and multilayer networks
• Activation functions (ReLU, sigmoid, tanh, Swish)
• Loss functions and optimization
• Convolutional Neural Networks (LeNet, AlexNet, VGG, ResNet)
• Recurrent Neural Networks and LSTM
• Transformers and self-attention
• Training best practices

Prerequisites: Linear algebra, calculus, Python programming
Level: Intermediate to Advanced
Duration: 3-4 weeks (graduate course)

---

Reinforcement Learning

📂 02-deep-learning/reinforcement-learning/

Learning Objectives:

• Formulate problems as Markov Decision Processes
• Derive and apply Bellman equations
• Implement value iteration and policy iteration
• Understand Monte Carlo and TD learning methods
• Build Q-learning and SARSA agents
• Apply function approximation with neural networks
• Implement modern deep RL algorithms (DQN, PPO, A3C)
• Design multi-agent systems

Topics Covered:

• Markov Decision Processes and dynamic programming
• Monte Carlo methods
• Temporal Difference learning (SARSA, Q-learning)
• Function approximation and deep Q-networks
• Policy gradient methods (REINFORCE, Actor-Critic, PPO)
• Multi-agent reinforcement learning
• Applications (games, robotics, resource allocation)

Prerequisites: Probability, linear algebra, Python
Level: Advanced
Duration: 4-5 weeks (graduate course)

---

🔷 Machine Learning Theory & Practice

Statistical Learning Theory

📂 01-foundations/statistical-modeling/

Learning Objectives:

• Understand bias-variance tradeoff
• Master regularization techniques (Ridge, Lasso, Elastic Net)
• Apply cross-validation and model selection
• Implement ensemble methods (bagging, boosting, stacking)
• Understand kernel methods and SVMs
• Perform dimensionality reduction (PCA, t-SNE, UMAP)
• Evaluate models using appropriate metrics

Topics Covered:

• Supervised learning fundamentals
• Linear and logistic regression
• Regularization and model selection
• Tree-based methods (CART, Random Forests, XGBoost)
• Support Vector Machines
• Gaussian Processes
• Model evaluation and validation

Prerequisites: Statistics, linear algebra, programming
Level: Intermediate
Duration: 4-5 weeks

---

Feature Engineering

📂 01-foundations/feature-engineering/

Learning Objectives:

• Design effective feature engineering pipelines
• Handle missing data with advanced imputation techniques
• Encode categorical variables appropriately
• Create polynomial and interaction features
• Apply feature scaling and normalization
• Perform feature selection using multiple methods
• Build end-to-end ML pipelines with scikit-learn

Topics Covered:

• Missing value imputation (mean, median, KNN, MICE)
• Categorical encoding (one-hot, ordinal, target, entity embeddings)
• Feature scaling (standard, min-max, robust)
• Polynomial features and interactions
• Feature selection (filter, wrapper, embedded methods)
• Dimensionality reduction
• Pipeline construction

Prerequisites: Basic Python, pandas, scikit-learn
Level: Beginner to Intermediate
Duration: 2-3 weeks

---

Explainable AI & Model Interpretability

📂 06-advanced-topics/explainable-ai/

Learning Objectives:

• Understand the interpretability-accuracy tradeoff
• Explain model predictions using SHAP values
• Apply LIME for local explanations
• Compute and interpret permutation importance
• Visualize partial dependence and ICE plots
• Detect and mitigate algorithmic bias
• Implement fairness metrics and constraints
• Use modern XAI tools (SHAP, LIME, InterpretML)

Topics Covered:

• Global vs local explanations
• Model-agnostic methods (SHAP, LIME, permutation importance)
• Model-specific interpretability (linear models, trees, neural networks)
• Attention mechanisms and gradient-based explanations
• Algorithmic fairness and bias detection
• Fairness definitions and impossibility results
• Practical implementation with Python tools

Prerequisites: Machine learning basics, Python
Level: Intermediate to Advanced
Duration: 2-3 weeks

---

🔷 Bayesian Methods & MCMC

Advanced MCMC Methods

📂 03-bayesian-methods/mcmc/

Learning Objectives:

• Understand Bayesian inference and posterior distributions
• Derive Metropolis-Hastings acceptance probability
• Implement MCMC algorithms from scratch
• Apply Hamiltonian Monte Carlo for efficient sampling
• Use No-U-Turn Sampler (NUTS) for automatic tuning
• Diagnose convergence using R-hat and ESS
• Apply MCMC to real Bayesian models

Topics Covered:

• Bayesian inference fundamentals
• Metropolis-Hastings algorithm
• Hamiltonian Monte Carlo and leapfrog integration
• No-U-Turn Sampler (NUTS)
• Convergence diagnostics (trace plots, R-hat, ESS)
• Applications (Bayesian regression, hierarchical models)

Prerequisites: Probability theory, calculus, Python
Level: Advanced
Duration: 3-4 weeks
Code: Complete Python implementations (8,000+ lines)

---

Bayesian Machine Learning

📂 03-bayesian-methods/bayesian-machine-learning/

Learning Objectives:

• Apply Bayesian inference to machine learning problems
• Build Bayesian linear and logistic regression models
• Implement Gaussian Processes for regression
• Understand Bayesian neural networks
• Perform approximate inference (VI, EP)
• Apply Bayesian optimization for hyperparameter tuning
• Quantify predictive uncertainty

Topics Covered:

• Bayesian linear regression
• Gaussian Processes
• Bayesian neural networks
• Variational inference
• Bayesian optimization
• Uncertainty quantification

Prerequisites: Bayesian statistics, machine learning, Python
Level: Advanced
Duration: 3-4 weeks

---

🔷 Causal Inference & Econometrics

Causal Inference

📂 04-causal-inference/causal-inference-fundamentals/

Learning Objectives:

• Understand potential outcomes framework
• Draw and interpret causal DAGs
• Implement Instrumental Variables (IV/2SLS)
• Apply Regression Discontinuity Design
• Use Difference-in-Differences methods
• Estimate propensity scores and perform matching
• Apply synthetic control methods
• Identify and address confounding

Topics Covered:

• Potential outcomes and causal graphs
• Instrumental Variables and weak instruments
• Regression Discontinuity (sharp and fuzzy)
• Difference-in-Differences and event studies
• Propensity score methods
• Synthetic controls
• Modern methods (Callaway-Sant'Anna, Sun-Abraham)

Prerequisites: Statistics, econometrics, R or Python
Level: Advanced
Duration: 4-5 weeks
Code: Python & R implementations (11,000+ lines)

---

🔷 Time Series & Forecasting

Time Series Analysis

📂 05-time-series/time-series-forecasting/

Learning Objectives:

• Analyze time series components (trend, seasonality)
• Test for and achieve stationarity
• Build ARIMA and SARIMA models
• Implement VAR models for multivariate series
• Apply state space models and Kalman filter
• Use LSTM and Transformers for forecasting
• Evaluate forecasting accuracy
• Apply hybrid methods (Prophet, N-BEATS)

Topics Covered:

• Stationarity and unit root tests
• ARMA, ARIMA, SARIMA models
• Vector Autoregression (VAR)
• State space models and Kalman filter
• Forecasting and evaluation
• Deep learning for time series (LSTM, GRU)
• Transformer models (TFT, Autoformer, Informer)
• Hybrid approaches (ES-RNN, N-BEATS, Prophet)

Prerequisites: Statistics, linear algebra, Python
Level: Intermediate to Advanced
Duration: 3-4 weeks

---

🔷 Optimization & Computational Methods

Optimization for Data Science

📂 01-foundations/optimization/

Learning Objectives:

• Formulate optimization problems
• Understand convexity and its implications
• Derive and apply KKT conditions
• Implement gradient descent variants
• Apply momentum and adaptive methods (Adam, AdamW)
• Solve constrained optimization problems
• Use evolutionary algorithms for black-box optimization
• Apply Bayesian optimization for hyperparameter tuning
• Optimize neural network training

Topics Covered:

• Convex optimization fundamentals
• Gradient descent (batch, SGD, mini-batch)
• Momentum methods and Nesterov acceleration
• Adaptive learning rates (AdaGrad, RMSProp, Adam)
• Constrained optimization (Lagrangian, KKT, penalties)
• Evolutionary algorithms (GA, ES, PSO, CMA-ES)
• Bayesian optimization
• Multi-objective optimization

Prerequisites: Calculus, linear algebra, Python
Level: Intermediate to Advanced
Duration: 3-4 weeks

---

🔷 Applied Data Science

A/B Testing & Experimentation

📂 04-causal-inference/ab-testing/

Learning Objectives:

• Design statistically rigorous A/B tests
• Calculate required sample sizes
• Perform hypothesis testing correctly
• Control for multiple comparisons
• Understand statistical power and effect sizes
• Apply sequential testing methods
• Analyze experimental results
• Avoid common pitfalls (peeking, p-hacking)

Topics Covered:

• Experimental design
• Hypothesis testing and p-values
• Sample size calculations
• Multiple testing corrections
• Bayesian A/B testing
• Sequential analysis
• Common pitfalls and best practices

Prerequisites: Statistics, probability
Level: Intermediate
Duration: 1-2 weeks

---

🏗️ Repository Structure

academic-presentations/
├── README.md                           # This file
├── CONTRIBUTING.md                     # Contribution guidelines
├── CHANGELOG.md                        # Version history
├── LICENSE                            # CC BY-SA 4.0 for content
│
├── .github/                           # 🤖 GitHub Actions automation
│   ├── workflows/
│   │   ├── compile-latex.yml         # Auto-compile PDFs
│   │   ├── check-links.yml           # Verify all URLs
│   │   └── generate-previews.yml     # Create PDF previews
│   ├── dependabot.yml                # Dependency updates
│   └── markdown-link-check-config.json
│
├── shared/                            # 🔄 Shared resources
│   ├── theme/                        # 🎨 Professional LaTeX theme
│   │   ├── esmad_beamer_theme.sty   # Custom Beamer theme
│   │   ├── esmad_beamer_theme_highcontrast.sty
│   │   ├── STYLE_GUIDE.md           # Theme documentation
│   │   └── template_presentation.tex # Example template
│   └── bibliographies/               # 📚 Reference libraries (140+ papers)
│       ├── mcmc_references.bib      # MCMC methods (30+ refs)
│       ├── causal_inference_references.bib # Causal inference (50+ refs)
│       └── statistical_learning_references.bib # ML/Stats (60+ refs)
│
├── 00-programming-fundamentals/      # 💻 Programming Basics
│   └── r-programming/
│       └── presentation/
│           └── R_programming.tex
│
├── 01-foundations/                   # 📊 Core Foundations
│   ├── statistical-modeling/
│   │   └── presentation/            # Statistical Learning Theory
│   ├── feature-engineering/
│   │   └── presentation/            # Feature Engineering
│   ├── pca/
│   │   └── presentation/            # Principal Component Analysis
│   └── optimization/
│       └── presentation/            # Optimization for Data Science
│
├── 02-deep-learning/                 # 🧠 Deep Learning
│   ├── deep-learning-fundamentals/
│   │   └── presentation/            # Deep Learning Fundamentals
│   └── reinforcement-learning/
│       └── presentation/            # Reinforcement Learning
│
├── 03-bayesian-methods/              # 🎲 Bayesian Statistics
│   ├── mcmc/
│   │   ├── presentation/            # MCMC Methods
│   │   └── exercises/               # MCMC Exercises
│   └── bayesian-machine-learning/
│       └── presentation/            # Bayesian ML
│
├── 04-causal-inference/              # ⚖️ Causal Methods
│   ├── causal-inference-fundamentals/
│   │   ├── presentation/            # Causal Inference Fundamentals
│   │   └── exercises/               # Causal Inference Exercises
│   └── ab-testing/
│       └── presentation/            # A/B Testing & Experimentation
│
├── 05-time-series/                   # ⏱️ Time Series
│   └── time-series-forecasting/
│       └── presentation/            # Time Series Analysis
│
├── 06-advanced-topics/               # 🔬 Advanced Topics
│   ├── explainable-ai/
│   │   └── presentation/            # Explainable AI
│   └── computer-science/
│       └── presentation/            # OOP & Streaming Pipelines
│
├── 07-capstone-projects/             # 🎓 Projects
│   ├── industry-focus/              # Industry applications
│   ├── project-guides/              # Project guidelines
│   └── prerequisites/               # Prerequisites
│
└── 08-data-science-applications-course/  # 🎯 Applied Course
    ├── presentation/                # Full course materials
    └── assessments/                 # Course assessments

🎨 Professional Theme & Styling

All presentations use the ESMAD Beamer Theme for consistent, professional appearance:

Theme Features

✅ Professional color palette (ESMAD Blue, accents)
✅ Custom environments (theorems, definitions, examples, alerts)
✅ Mathematical notation helpers (\Normal, \E, \Var, etc.)
✅ Code listing styles with syntax highlighting
✅ Author information with ORCID integration
✅ Slide templates (title, TOC, contact, references)

Usage

\documentclass[aspectratio=169]{beamer}
\usepackage{../../../shared/theme/esmad_beamer_theme}

% Author info
\authorname{Your Name}
\authoremail{your.email@university.edu}
\authororcid{0000-0000-0000-0000}

\title{Your Presentation}
\date{\today}

\begin{document}
\begin{frame}
  \titlepage
\end{frame}

% Your content...

\contactslide
\end{document}

See shared/theme/STYLE_GUIDE.md for complete documentation.

🔧 Getting Started

Prerequisites

LaTeX Distribution:

# Ubuntu/Debian
sudo apt-get install texlive-full

# macOS
brew install --cask mactex

# Windows
# Download and install MiKTeX or TeX Live

Python Environment (for code examples):

pip install numpy scipy matplotlib seaborn pandas scikit-learn statsmodels
pip install torch tensorflow  # For deep learning examples
pip install shap lime  # For XAI examples

R Environment (for R examples):

install.packages(c(
  "AER", "rdrobust", "fixest", "did",  # Causal inference
  "caret", "recipes", "mice",           # Feature engineering
  "forecast", "vars", "fable"           # Time series
))

Compiling Presentations

Manual compilation:

cd 02-deep-learning/deep-learning-fundamentals/presentation/
pdflatex deep_learning_beamer.tex
pdflatex deep_learning_beamer.tex  # Run twice for references

Using latexmk (recommended):

cd 02-deep-learning/reinforcement-learning/presentation/
latexmk -pdf rl_beamer.tex

Automated compilation:

• Push to GitHub → GitHub Actions automatically compiles all PDFs
• Download compiled PDFs from Actions artifacts or Releases

Running Code Examples

Python:

# MCMC examples (if code/ directory exists with implementations)
# Example references are embedded in presentation materials

# Exercises and assessments
cd 03-bayesian-methods/mcmc/exercises/
pdflatex mcmc_exercises.tex

Exercises:

# MCMC exercises
cd 03-bayesian-methods/mcmc/exercises/
pdflatex mcmc_exercises.tex

# Causal inference exercises
cd 04-causal-inference/causal-inference-fundamentals/exercises/
pdflatex causal_inference_exercises.tex

📖 For Students

Recommended Learning Paths

Path 1: Machine Learning Fundamentals

1. Statistical Learning (4 weeks)
2. Feature Engineering (2 weeks)
3. Optimization (3 weeks)
4. Explainable AI (2 weeks)

Path 2: Deep Learning Specialization

1. Deep Learning Fundamentals (4 weeks)
2. Optimization (focus on neural networks)
3. Reinforcement Learning (4 weeks)
4. Time Series Analysis (focus on deep methods)

Path 3: Causal & Bayesian Methods

1. Causal Inference (5 weeks)
2. Bayesian ML (4 weeks)
3. MCMC Methods (3 weeks)
4. A/B Testing (2 weeks)

Study Tips

• 📚 Start with slides to understand concepts
• 💻 Run code examples to see methods in action
• 📝 Complete exercises to test understanding
• 📖 Read references for deeper knowledge
• 🤝 Join discussions (create GitHub issues)

👨‍🏫 For Educators

Course Integration

These materials can be integrated into:

• Graduate courses in Data Science/Statistics/CS
• Professional training programs
• Workshop series
• Seminar courses

Customization

1. Fork this repository
2. Customize presentations for your needs
3. Add your own examples and exercises
4. Maintain attribution (CC BY-SA 4.0)

Assessment Resources

Use the materials in assessments/:

• Quizzes for each topic
• Midterm and final exams
• Grading rubrics
• Project ideas

🔬 For Researchers

Citation

If you use these materials in your research or teaching, please cite:

@misc{ribeiro2025academic,
  author = {Ribeiro, Diogo},
  title = {Academic Presentations: Comprehensive Data Science Course Materials},
  year = {2025},
  publisher = {GitHub},
  url = {https://github.com/diogoribeiro7/academic-presentations},
  note = {ESMAD \& Mysense.ai}
}

Using the Bibliographies

All presentations reference comprehensive BibTeX files:

\usepackage[backend=bibtex]{biblatex}
\addbibresource{../../../shared/bibliographies/mcmc_references.bib}

% In document
\cite{metropolis1953}
\cite{hoffman2014}

% At end
\printbibliography

Available:

• shared/bibliographies/mcmc_references.bib: 30+ MCMC papers
• shared/bibliographies/causal_inference_references.bib: 50+ causal inference papers
• shared/bibliographies/statistical_learning_references.bib: 60+ ML/stats papers

All include DOIs for easy access.

🤝 Contributing

We welcome contributions! See CONTRIBUTING.md for guidelines.

How to Contribute

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Test compilation and code
5. Submit a pull request

Contribution Types

• 🐛 Fix errors in presentations
• 📚 Add new presentations
• 💡 Improve existing content
• 📖 Enhance documentation
• 🧪 Add code examples
• 📝 Create exercises
• 🎨 Improve theme/styling

🤖 Automation & CI/CD

GitHub Actions Workflows

• compile-latex.yml: Auto-compiles all LaTeX on push
• check-links.yml: Verifies all URLs and DOIs weekly
• generate-previews.yml: Creates PDF preview gallery
• dependabot.yml: Keeps dependencies updated

PDF Preview Gallery

View slide previews at: https://diogoribeiro7.github.io/academic-presentations/

📊 Repository Statistics

• 📚 15+ comprehensive presentations
• 💻 27,000+ lines of code (Python & R)
• 📖 140+ curated references with DOIs
• 📝 14 pages of exercises (2 comprehensive problem sets)
• 🎨 1 professional LaTeX theme with full documentation
• 🤖 Fully automated PDF compilation and testing

📄 License

Content (Presentations & Exercises)

Licensed under Creative Commons Attribution-ShareAlike 4.0 International

You are free to:

• ✅ Share — copy and redistribute
• ✅ Adapt — remix, transform, and build upon

Under the terms:

• 📝 Attribution required
• 🔄 ShareAlike for derivatives

Code

Code examples licensed under MIT License

📞 Contact & Collaboration

Professional Inquiries

• Email: dfr@esmad.ipp.pt
• Institution: ESMAD - Escola Superior de Média Arte e Design
• Company: Mysense.ai (Lead Data Scientist)
• ORCID: 0009-0001-2022-7072

Research Interests

• Markov Chain Monte Carlo and Bayesian computation
• Machine learning and deep learning
• Causal inference and econometrics
• Financial risk modeling
• Time series analysis and forecasting

Collaboration Opportunities

• 🎓 Guest lectures and workshops
• 🏢 Corporate training programs
• 🔬 Research collaborations
• 📝 Joint publications
• 🌐 Conference presentations

🌟 Acknowledgments

• ESMAD for institutional support
• Mysense.ai for industry applications and insights
• Students and colleagues for valuable feedback
• Open source community for tools and inspiration
• Academic community for rigorous peer review

📈 Version History

See CHANGELOG.md for detailed version history.

---

Last Updated: January 2025
Repository Maintainer: Diogo Ribeiro
Status: ✅ Actively maintained
Latest Release: View releases