Home/ Majors directory/Computer Science/Neural Network Technologies and Their Applications
Neural Network Technologies and Their Applications
Major: Computer Science
Code of subject: 6.122.00.M.079
Credits: 6.00
Department: Artificial Intelligence Systems
Lecturer: O. Gurbych
Semester: 6 семестр
Mode of study: денна
Learning outcomes: Software Engineering
Languages: Python or R
Tools: git, Jupyter Notebook, Colab, Gradient, or similar
Implement function, class
Refactoring
Debugging
Import library/module
Documentation
Data Manipulation
Libraries: PyTorch, TensorFlow, fast.ai, MXNet, or similar
Data Preprocessing
Libraries: pandas, PySpark, Dask, or similar
Imputing missing values
Drop duplicates
MinMax, Normalization, Standardization, etc.
Mean, median, mode
Correlation analysis
Data Visualization
Libraries: matplotlib, seaborn, plotly, ggplot or similar
Linear Algebra
Libraries: numpy, scipy
Scalars
Vectors
Matrices
Tensors, Tensor Arithmetic
Reduction
Dot Products
Matrix-Matrix Multiplication
Norms
Calculus
Libraries: numpy, scipy
Differentiation and Derivatives
Partial Derivatives
Gradients
Chain Rule
Probability Theory
Basic Probability Theory
Dealing with Multiple Random Variables
Expectation and Variance
Bayes Theorem
Machine Learning
Libraries: sklearn
Models: Linear, Tree Ensembles, Probabilistic, etc.
Required prior and related subjects: Software Engineering
Languages: Python or R
Tools: git, Jupyter Notebook, Colab, Gradient, or similar
Implement function, class
Refactoring
Debugging
Import library/module
Documentation
Data Manipulation
Libraries: PyTorch, TensorFlow, fast.ai, MXNet, or similar
Data Preprocessing
Libraries: pandas, PySpark, Dask, or similar
Imputing missing values
Drop duplicates
MinMax, Normalization, Standardization, etc.
Mean, median, mode
Correlation analysis
Data Visualization
Libraries: matplotlib, seaborn, plotly, ggplot or similar
Linear Algebra
Libraries: numpy, scipy
Scalars
Vectors
Matrices
Tensors, Tensor Arithmetic
Reduction
Dot Products
Matrix-Matrix Multiplication
Norms
Calculus
Libraries: numpy, scipy
Differentiation and Derivatives
Partial Derivatives
Gradients
Chain Rule
Probability Theory
Basic Probability Theory
Dealing with Multiple Random Variables
Expectation and Variance
Bayes Theorem
Machine Learning
Libraries: sklearn
Models: Linear, Tree Ensembles, Probabilistic, etc.
Summary of the subject: Introduction to Deep Learning
What Deep Learning Is
Kinds of Deep Learning
Deep Learning Projects - Generalized Pipeline
Key Components
From Biological to Artificial Neurons
Tabular / DNN
Linear Neural Networks
Multilayer Feed-Forward Perceptrons
Regression & MSE
Binary Classification & LogLoss
Multiclass Classification & Categorical Cross-Entropy
Training Neural Networks
Model Selection
Underfitting & Overfitting
Weights Initialization: Glorot, He, Xavier
Nonsaturating Activation Functions
Regression Loss Functions
Classification Loss Functions
Reconstruction Loss Functions
Optimizers: SGD, Momentum, Nesterov, AdaGrad, RMSProp, Adam, Adam
Regularization: L1, L2, Dropout, Monte-Carlo Dropout, Max-Norm
Saving and restoring model
Callbacks: Early Stopping, LR Decay on Plateau, Snapshots
TensorBoard
Hyperparameters Optimization
Number of Hidden Layers
Number of Neurons per Hidden Layer
Learning Rate
Batch Size
Other Hyperparameters
Vision / CNN
1D, 2D, 3D Conv layers, filters, padding, stride, pooling, feature maps
Image Classification
CNN Architectures
ResNetV2
DenseNet
MobileNetV3
Xception
SENet
ImageNet
Using pre-trained models: fine-tuning vs transfer learning
Object Detection - review 2021
YOLOv5
SSD
FRCNN
OFA
Swin Transformer
Semantic Segmentation - Hands-On-Machine-Learning
UNet
PSPNet
Language / NLP / RNN
Text Preprocessing
Tokenization
Numerization
Word Embeddings
word2vec
GloVe
Recurrent Neural Networks (RNNs)
GRU
LSTM
Attention Mechanisms
Transformers
BERT
GPT-2, GPT-3
Recommender Systems
Generative Adversarial Networks (GANs)
Deep Reinforcement Learning
DQN
A2C
MLOps
ML Infrastructure & Operations
Deployment & Monitoring
Assessment methods and criteria: project - 100
Recommended books: Dive into Deep Learning by Alex Smola, 2021
The hundred-page machine learning book by Andriy Burkov, 2020
Machine Learning Engineering by Andriy Burkov, 2020
Deep Learning Book by Ian Goodfellow, Yoshua Bengio; MIT, 2017
ML Yearning book, by Andrew Ng; 2019
Stanford CS 329P: Practical Machine Learning
Stanford CS 329S: Machine Learning Systems Design; Stanford, Winter 2021
Stanford CS229: Machine Learning; Stanford, Spring 2021
MIT 6.S191: Introduction to Deep learning, 2020
MIT Deep Learning and Artificial Intelligence Lectures, 2019-2020