Deep-Learning-in-Hebrew

למידת מכונה ולמידה עמוקה בעברית

Add a star if the repository helped you 😊

For any issue please contact us at Avrahamsapir1@gmail.com.

📓 Notebooks & Full Book (NEW!)

This repository includes Jupyter notebooks that accompany the book and reflect its chapter-by-chapter structure.

The full content (complete book PDFs + all notebooks + bonus material) is available via the project website.

People

Authors:

• Avraham Raviv

• Mike Erlihson

Chapter Authors:

• David Ben Attar

• Gal Peretz

• Hadar Sharvit

• Jeremy Rutman

• Maya Rapaport

• Nava (Reinitz) Leibovich

• Nitzan Madar

• Or Avrahami

• Or Shemesh

• Ron Levy

• Uri Almog

Contributors:

• Avi Caciularu

• Avshalom Dayan

• Nitzan Madar

• Rachel Wities

• Tomer Caspi

Citation

If you find this book useful in your research work, please consider citing:

@InProceedings{MDLH,
author = {Raviv, Avraham and Erlihson, Mike},
booktitle = {Machine and Deep learning in Hebrew},
year = {2021}
}

---

Table of Content

1. Introduction to Machine Learning

1.1 What is Machine Learning?

• 1.1.1. The Basic Concept

• 1.1.2. Data, Tasks and Learning

1.2. Applied Math

• 1.2.1. Linear Algebra

• 1.2.2. Calculus

• 1.2.3. Probability

2. Machine Learning Algorithms

2.1. Supervised Learning Algorithms

• 2.1.1. Support Vector Machines (SVM)

• 2.1.2. Naïve Bayes

• 2.1.3. K-nearest neighbors (K-NN)

• 2.1.4. Quadratic\Linear Discriminant Analysis (QDA\LDA)

• 2.1.5. Decision Trees

2.2. [Unsupervised Learning Algorithms]

• 2.2.1. K-means

• 2.2.2. Mixture Models

• 2.2.3. Expectation–maximization (EM)

• 2.2.4. Hierarchical Clustering

• 2.2.5. Local Outlier Factor

2.3. [Dimensionally Reduction]

• 2.3.1. Principal Components Analysis (PCA)

• 2.3.2. t-distributed Stochastic Neighbor Embedding (t-SNE)

• 2.3.3. Uniform Manifold Approximation and Projection (UMAP)

2.4. [Ensemble Learning]

• 2.4.1. Introduction to Ensemble Learning

• 2.4.2. Bagging

• 2.4.3. Boosting

3. Linear Neural Networks (Regression problems)

3.1. Linear Regression

• 3.1.1. The Basic Concept

• 3.1.2. Gradient Descent

• 3.1.3. Regularization and Cross Validation

• 3.1.4. Linear Regression as Classifier

3.2. Softmax Regression

• 3.2.1. Logistic Regression

• 3.2.2. Cross Entropy and Gradient Descent

• 3.2.3. Optimization

• 3.2.4. SoftMax Regression – Multiclass Logistic Regression

• 3.2.5. SoftMax Regression as Neural Network

4. Deep Neural Networks

4.1. MLP – Multilayer Perceptrons

• 4.1.1. From a Single Neuron to Deep Neural Network

• 4.1.2. Activation Function

• 4.1.3. Xor

4.2. Computational Graphs and Propagation

• 4.2.1. Computational Graphs

• 4.2.2. Forward and Backward propagation

• 4.2.3. Back Propagation and Stochastic Gradient Descent

4.3. Optimization

• 4.3.1. Data Normalization

• 4.3.2. Weight Initialization

• 4.3.3. Batch Normalization

• 4.3.4. Mini Batch

• 4.3.5. Gradient Descent Optimization Algorithms

4.4. Generalization

• 4.4.1. Regularization

• 4.4.2. Weight Decay

• 4.4.3. Model Ensembles and Drop Out

• 4.4.4. Data Augmentation

5. Convolutional Neural Networks

5.1. Convolutional Layers

• 5.1.1. From Fully-Connected Layers to Convolutions

• 5.1.2. Padding, Stride and Dilation

• 5.1.3. Pooling

• 5.1.4. Training

• 5.1.5. Convolutional Neural Networks (LeNet)

5.2. CNN Architectures

• 5.2.1. AlexNet

• 5.2.2. VGG

• 5.2.3. GoogleNet

• 5.2.4. Residual Networks (ResNet)

• 5.2.5. Densely Connected Networks (DenseNet)

• 5.2.6. U-Net

• 5.2.7. Transfer Learning

6. Recurrent Neural Networks

6.1. Sequence Models

• 6.1.1. Recurrent Neural Networks

• 6.1.2. Learning Parameters

6.2. RNN Architectures

• 6.2.1. Long Short-Term Memory (LSTM)

• 6.2.2. Gated Recurrent Units (GRU)

• 6.2.3. Deep RNN

• 6.2.4. Bidirectional RNN

• 6.2.5. Sequence to Sequence Learning

7. Deep Generative Models

7.1. Variational AutoEncoder (VAE)

• 7.1.1. Dimensionality Reduction

• 7.1.2. Autoencoders (AE)

• 7.1.3. Variational AutoEncoders (VAE)

7.2. Generative Adversarial Networks (GANs)

• 7.2.1. Generator and Discriminator

• 7.2.2. DCGAN

• 7.2.3. Conditional GAN (cGAN)

• 7.2.4. Pix2Pix

• 7.2.5. CycleGAN

• 7.2.6. Progressively Growing (ProGAN)

• 7.2.7. StyleGAN

• 7.2.8. Wasserstein GAN

7.3. Auto-Regressive Generative Models

• 7.3.1. PixelRNN

• 7.3.2. PixelCNN

• 7.3.3. Gated PixelCNN

• 7.3.4. PixelCNN++

8. Attention Mechanism

8.1. Sequence to Sequence Learning and Attention

• 8.1.1. Attention in Seq2Seq Models

• 8.1.2. Bahdanau Attention and Luong Attention

8.2. Transformer

• 8.2.1. Positional Encoding

• 8.2.2. Self-Attention Layer

• 8.2.3. Multi Head Attention

• 8.2.4. Transformer End to End

• 8.2.5. Transformer Applications

9. Computer Vision

9.1. Object Detection

• 9.1.1. Introduction to Object Detection

• 9.1.2. R-CNN

• 9.1.3. You Only Look Once (YOLO)

• 9.1.4. Single Shot Detector (SSD)

• 9.1.5 Spatial Pyramid Pooling (SPP-net)

• 9.1.6. Feature Pyramid Networks

• 9.1.7. Deformable Convolutional Networks

• 9.1.8. DE:TR: Object Detection with Transformers

9.2. Segmentation

• 9.2.1. Semantic Segmentation Vs. Instance Segmentation

• 9.2.2. SegNet neural network

• 9.2.3. Atrous Convolutions

• 9.2.4. Atrous Spatial Pyramidal Pooling

• 9.2.5. Conditional Random Fields usage for improving final output

• 9.2.6. See More Than Once -- Kernel-Sharing Atrous Convolution

9.3. Face Recognition and Pose Estimation

• 9.3.1. Face Recognition

• 9.3.2. Pose Estimation

9.5. Few-Shot Learning

• 9.5.1. The Problem

• 9.5.2 Metric Learning

• 9.5.3. Meta-Learning (Learning-to-Learn)

• 9.5.4. Data Augmentation

• 9.5.5. Zero-Shot Learning

10. Natural Language Process

10.1. Language Models and Word Representation

• 10.1.1. Basic Language Models

• 10.1.2. Word Representation (Vectors) and Word Embeddings

• 10.1.3. COntextual Embeddings

11. Reinforcement Learning

11.1. Introduction to RL

• 11.1.1. Markov Decision Process (MDP) and RL

• 11.1.2. Planning

• 11.1.3. Learning Algorithms

11.2. Model Free Prediction

• 11.2.1. Monte-Carlo (MC) Policy Evaluation

• 11.2.2. Temporal Difference (TD) – Bootstrapping

• 11.2.3. TD(λ)

11.3. Model Free Control

• 11.3.1. SARSA - on-policy TD control

• 11.3.2. Q-Learning

• 11.3.3. Function Approximation

• 11.3.4. Policy-Based RL

• 11.3.5. Actor-Critic

11.4. Model Based Control

• 11.4.1. Known Model – Dyna algorithm

• 11.4.2. Known Model – Tree Search

• 11.4.3. Planning for Continuous Action Space

11.5. Exploration and Exploitation

• 11.5.1. N-armed bandits

• 11.5.2. Full MDP

11.6. Learning From an Expert

• 11.6.1. Imitation Learning

• 11.6.2. Inverse RL

11.7. Partially Observed Markov Decision Process (POMDP)

12. Graph Neural Networks

12.1. Introduction to Graphs

• 12.1.1. Represent Data as a Graph

• 12.1.2. Tasks on Graphs

• 12.1.3. The challenge of learning graphs

---

References

• Stanford cs231
• Machine Learning - Andrew Ng
• Dive into Deep Learning
• Deep Learning Book

כל הזכויות שמורות Ⓒ