## University of Southern California

### Electrical Engineering

- EE 682: Law and Intellectual Property for Engineers (Fall 2020)

Outline

- American legal system; Intellectual property
- Torts I: Intentional harm and negligence
- Torts II: Strict liability; Patent and copyright infringement
- Criminal law and procedure; Theft of trade secrets
- Contracts I: Making and breaking promises
- Contracts II: Third-party liability; Assignments and licenses
- Law of property; Estates; Transfers
- Civil procedure and federal courts; Standing
- Constitutional law; Congressional authority; State vs. federal law
- Evidence; Standards of proof; Relevance
- Corporations; Cyberlaw
- Patents; Drafting claims
- Copyright; Trade secrets; Trademarks

- Introduction to Mathematics of Data; Sample Applications; Optimization Basics
- Optimization for Modern Data Analysis I: First Order Methods, Accelerated Schemes
- Optimization for Modern Data Analysis II: Sub-Gradients and Non-Smooth Optimization, Incremental and Stochastic Schemes
- Basics of Concentration of Measure and High Dimensional Probability
- Non-Asymptotic Random Matrix Theory and Matrix Concentration
- Dimension Reduction, Sketching, and Applications
- Fast and Randomized Methods for Numerical Linear Algebra
- Clustering I: Matrix Perturbation Theory
- Clustering II: Spectral Algorithms, Application in Community Detection
- Linear Inverse Problems I: Compressive Sensing and Sparsity, Recommender Systems, Matrix Completion and Low-Rank Modeling
- Linear Inverse Problems II: Recovery of ne-scale Data from coarse-scale Measurements: Applications in Deblurring, Fluorescence Microscopy, Wireless Communications, Medical Imaging and Computer Vision
- Modern Theory of Linear Inverse Problems; Iterative Algorithms and Non-Convex Optimization; Phase Retrieval and Computational Imaging
- Discrete and Submodular Optimization and Learning
- Learning Representations; Sparse Coding; Word Embeddings
- Kernel Methods; “Shallow" and “Deep" Learning

- Introduction to Machine Learning
- Key Issues and Concepts in Machine Learning
- Multidimensional Regression: Linear Regression, Maximum-Likelihood and MAP Estimation, Ridge Regression, Bayesian regression; Learning Linear and Non-Linear Relationships
- Review of Convexity and Optimization
- Logistic Regression
- Feasibility of Learning: Deterministic and Statistical Views; Hoeffding Inequality; Inductive Bias
- Complexity of Learning I: Generalization; Estimation of Error on New Data; Implications in Dataset Usage
- Complexity of Learning II: Bias-Variance Decomposition; Learning Curves; Overfitting
- Regularization; Feature Reduction; Sparsity
- Model Selection and Validation
- Boosting Techniques and Decision Trees
- Kernel Methods
- Semi-Supervised Learning for Classification
- Unsupervised Learning for Clustering: Statistical Techniques
- Unsupervised Learning for Clustering: Other Techniques

- Basic Concepts in Pattern Recognition; A Paradigm in Pattern Recognition
- Distribution-Free Classification I: Classifier Design; Discriminant Functions
- Distribution-Free Classification II: Training and Optimization for Supervised Learning; Perceptron; Support Vector Machines
- Statistical Classification I: Statistics are Known - Bayes Decision Theory
- Statistical Classification II: Statistics are Partially Known - Parameter Estimation: Maximum Likelihood, Maximum A Posteriori, Bayesian Estimation
- Statistical Classification III: Statistics are Unknown - Non-Parametric Techniques: Histogram, Parzen Windows, k-Nearest Neighbors
- Statistical Classification IV: Supervised Learning
- Validation and Cross-Validation; Feature Selection and Reduction
- Artificial Neural Networks

- Overview of Statistics; Probability Review
- Sampling Distributions
- Point Estimation
- Confidence Intervals
- Hypothesis Testing
- Tests for Probability Densities; Contingency Tables
- Sufficient Statistics; Cramer-Rao Bound; Ratio Hypothesis Tests
- Sequential Tests; Linear Regression; Heteroscedasticity
- Multiple Regression; Multicollinearity Diagnostics
- Model Building; Stepwise Regression; Statistical Process Control
- Other Regression Types; ANOVA
- Runs; Experimental Design; Bayesian Statistics
- Expectation-Maximization; Hierarchical Bayes and Gibbs Samplers; Non-Parametric/Robust Tools

- EE 503: Probability for Electrical and Computer Engineers (Fall 2016)

Outline

- Logic and Sets; Sigma Algebras; Probability Axioms
- Independence; Total Probability; Bayes' Theorem
- Combinatorics; Binomial Theorem; Limits of Sequences
- Poisson Theorem; Negative Binomial; Formal Reasoning
- Random Variables; Densities and Cumulative Distributions
- Expectations and Moments of Random Variables
- Covariance; Correlation; Uncertainty Principles
- Stochastic Convergence; Laws of Large Numbers
- Conditional Expectations; Maximum Likelihood Estimation
- Transformed Densities; Random Sampling; Entropy
- Central Limit Theorem and Applications; Confidence Intervals
- Financial Engineering; Introduction to Martingales and Markov Chains
- Markov Chains: Estimation
- Markov Chains and Queues: Advanced Applications

- EE 562: Random Processes in Engineering (Fall 2016)

Outline

- Definition of Random Processes: Random Variables, Random Vectors, Random Sequences, Random Waveforms, etc.
- Second Order Statistics: Properties of Correlation Functions
- Covariance Matrix Factorization; Eigenvalues - Eigenvectors; Causal Factoring and Whitening Concepts
- Gaussian Processes
- Simple Hypothesis Tests
- Linear Minimum-Mean-Square-Error Estimation; Orthogonality Principle
- Linear Operations on Random Processes; Convergence Concepts: Convolution, Integration, Differentiation
- Frequency Domain Analysis: Time Invariant Linear Operations
- Energy Spectra; Power Spectra; White Noise Approximations
- Linear Transformations of Wide-Sense Stationary Random Processes; Spectral Factorization; Applications
- Poisson Distributed Events in Time; Campbell’s Theorem
- Karhuenen-Loeve Expansions of Finite Intervals
- Narrowband Process Representations
- Time Averages; Ergodicity

### Computer Science

- CSCI 570: Analysis of Algorithms (Summer 2021)

Outline

- Introduction: Stable Matching, Asymptotic Notation, BFS, DFS
- Greedy Algorithms
- Heaps and Amortized Cost
- Minimum Spanning Trees
- Shortest Paths
- Divide and Conquer
- Dynamic Programming
- Network Flow: Max Flow
- Network Flow: Circulation
- NP-Completeness
- Approximation Algorithms
- Linear Programming

- CSCI 535: Multimodal Probabilistic Learning of Human Communication (Spring 2021)

Outline, Final Project

- Introduction and Communication Models
- Machine Learning: basic concepts
- Study Design, Evaluation and Analysis
- Affective messages and personality traits
- Vocal messages
- Virtual Humans
- Verbal and Conversational messages
- Multimodal behavior recognition
- Human-Robot Communication
- Multimodal Sentiment Analysis
- Dyadic and Multiparty Interactions
- Bias and Ethics

- Introduction and Probability Basics
- Linear Classifiers: Naive Bayes, Logistic Regression, Perceptron
- Nonlinear Classifiers, feed-forward neural networks, backpropagation, gradient descent
- POS tags, HMMs, search
- Parsing and Syntax I: treebanks, evaluation, CKY, grammar induction, PCFGs
- Parsing and Syntax II: dependencies, shift-reduce
- Evaluation, Annotation, Mechanical Turk
- Semantics: Word sense, PropBank, AMR, distributional and lexical semantics
- Language Models: n-gram, feed-forward, recurrent
- Machine Translation history, evaluation, statistical MT
- Neural Machine Translation, summarization, generation
- Transformers
- Large Contextualized Language Models (ElMo, BERT, GPT, etc.)
- Information Extraction: Entity/Relation, CRF, Events, zero-shot
- Blade Runner NLP / BERTology
- Text games and Reinforcement Learning
- Dialogue
- Power and Ethics

- Introduction to Computational Social Sciences
- Dictionary Methods
- Differential Language Analysis
- Latent Semantic Analysis
- Neural Networks
- Clinical and Cognitive Applications
- Ethics

- Introduction to Machine Learning Theory; Supervised Learning
- Online Learning: Winnow, Best Experts, Weighted Majority, Perceptron Algorithms
- Generic Bounds for Online Learning
- VC dimension and Sample Complexity
- PAC Learning Model; Online to PAC Conversion; Occam's Razor
- VC Dimension and Sample Complexity of PAC Learning
- Weak and Strong Learning
- Boosting
- PAC Learning with Noise: Random Classification Noise, Malicious Noise
- Statistical Query (SQ) Learning; Simulating SQ Queries in the Presence of Random Classification Noise
- Adaboost Algorithm and Analysis

- Machine Learning Review
- Loss Functions and Optimization; Feed Forward Neural Networks
- Convolutional Neural Networks (CNNs)
- Training Neural Networks
- CNN Architectures
- Deep Learning Software
- Recurrent Neural Networks
- Generative Adversarial Networks (GANs)
- Variational Autoencoders
- PixelRNN; PixelCNN
- Deep Reinforcement Learning
- InfoGAN; CycleGAN
- Attention Networks; Relational Networks; Memory Networks
- AlphaGo; AlphaGo Zero
- Imitation Learning

## National Technical University of Athens

*This is a subset of the advanced classes I took at NTUA which are relevant to Signal Processing, Artificial Intelligence, or Software Development.*

- Pattern Recognition with Emphasis on Speech Recognition
- Speech and Natural Language Processing
- Computer Vision
- Image and Video Analysis and Technology
- Digital Signal Processing
- Neural Networks and Intelligent Systems
- Knowledge Systems and Technologies
- Artificial Intelligence
- Software Engineering
- Algorithms and Complexity
- Operating Systems
- Databases
- Programming Languages I