Machine Learning from Scratch

This repository contains examples of popular machine learning algorithms implemented in Python with mathematics behind them being explained. Each algorithm has interactive Jupyter Notebook demo that allows you to play with training data, algorithms configurations and immediately see the results, charts and predictions right in your browser.

The purpose of this repository is not to implement machine learning algorithms by using 3^rd party library one-liners but rather to practice implementing these algorithms from scratch and get better understanding of the mathematics behind each algorithm.

Table of Contents

• Machine Learning
  • Supervised Learning
    • Regression
      • Linear Regression
      • Polynomial Regression
      • Lasso/Ridge Regression
      • Support Vector Machines
      • K nearest neighbors
    • Classification
      • Logistic Regression
      • Naive Bayes
      • Decision Tree
    • Ensemble Method
      • Bootstrap Aggregating
      • Random Forest
      • Boosting
      • Adaptative Boosting
      • Gradient Boosting
      • Stacking
  • Unsupervised Learning
    • Dimentional Reduction
      • Principal Component Analysis
      • Non-Negative Matrix Factorization
      • Singular Value Decomposition
      • Non-Negative Matrix Factorization
    • Clustering
      • K-Mean
      • DBSCAN
• Deep Learning
  • Perceptron
  • Artificial Neural Network
  • Convolutional Neural Network
  • Recurrent Neural Network
• Optimization Algorithms
  • Gradient Decent
  • Gradient Decent Check
  • Gradient Decent with Mini-Batch
  • Gradient Decent with Adam Optimization
  • Gradient Decent with Momentum Optimization
  • Newton's Raphson Method
• Paper Implement
  • LeNet
  • AlexNet
  • GoogLeNet
  • ResNet
• Deep Learning with Tensorflow
  • Regression
  • Classification
  • Complex Modelling using Functional API
  • AutoEncoder
  • Variational AutoEncoder
  • Generative Adversarial Networks
  • Attention
  • Transformer

Machine Learning

Supervised Learning

In supervised learning we have a set of training data as an input and a set of labels or correct answers for each training set as an output. Then we're training our model (machine learning algorithm parameters) to map the input to the output correctly (to do correct prediction). The ultimate purpose is to find such model parameters that will successfully continue correct input→output mapping (predictions) even for new input examples.

Regression

In regression problems we do real value predictions. Basically we try to draw a line/plane/n-dimensional plane along the training examples. In regression we deal with continuos as well as decreate data

🤖 Linear Regression

• 📗 Math | Linear Regression - theory and links for further readings
• ▶️ Demo | Univariate Linear Regression

🤖 Polynomial Regression

• 📗 Math | Polynomial Regression - theory and links for further readings
• ▶️ Demo | Univariate Polynomial Regression

🤖 Lasso/Ridge Regression

• ▶️ Demo | Lasso/Ridge Regression -Both regressions are explained with Neural Network. But you can apply in any algorithm also.

🤖 Support Vector Machines

SVM construct a hyper plane in high dimension which can be used for Classification , regression , or outlier detection.

• 📗 Math | SVM - theory and code
• ▶️ Demo | SVM Hard Margin - theory and code
• ▶️ Demo | SVM Soft Margin - theory and code
• ▶️ Demo | SVM Kernel Trick - theory and code
• ▶️ Demo | SVM Sequential Minimal Optimization( SMO ) - theory and code
• ▶️ Demo | SVM (Multi-class) - theory and code

🤖 K nearest neighbors

• KNN (K — Nearest Neighbors) is one of many (supervised learning) algorithms used in data mining and machine learning, it’s a classifier algorithm where the learning is based “how similar” is a data (a vector) from other

• 📗 Math | KNN - theory
• ▶️ Demo | KNN - code from scratch.

Classification

In classification problems we split input examples by certain characteristic.

_Usage examples: benign-malignent-data, wine-quality, MNIST handwritten.

🤖 Logistic Regression

• 📗 Math | Logistic Regression - theory and links for further readings
• ▶️ Demo | Logistic Regression (Linear Boundary)
• ▶️ Demo | Multivariate Logistic Regression | Wine-quality
• ▶️ Demo | Multivariate Logistic Regression | Benign-Malignent

🤖 Naive Bayes

• 📗 Math | Naive Bayes Classifier - theory and links for further readings
• ▶️ Demo | Bivariate Naive Bayes Classifier | Benign-Malignent

🤖 Decision Tree

• ▶️ Demo | Decision Tree | Wine-quality from scratch

• ▶️ Demo | Decision Tree | Multi-class

Ensemble Method

Ensemble learning is a machine learning paradigm where multiple models (often called “weak learners”) are trained to solve the same problem and combined to get better results. The main hypothesis is that when weak models are correctly combined we can obtain more accurate and/or robust models.

• 📗 Theory | Ensemble Learning

• 📗 Theory | Bootstrap Aggregating

• 📗 Theory | Random Forest

• ▶️ Demo | Random Forest

• ▶️ Demo | Random Forest (hypertune)

• 📗 Theory | Boosting

• ▶️ Demo | Adaptative Boosting

• 📗 Theory | Gradient Boosting

• ▶️ Demo | Gradient Boosting

• 📗 Theory | Stacking

• ▶️ Demo | eXtreme Gradient Boosting(XGBoost)

Unsupervised Learning

Unsupervised learning is a branch of machine learning that learns from test data that has not been labeled, classified or categorized. Instead of responding to feedback, unsupervised learning identifies commonalities in the data and reacts based on the presence or absence of such commonalities in each new piece of data.

Dimentional Reduction

In dimentional reduction we select 'K' features from given 'n' features by using some techniques.

🤖 Principal Component Analysis

• 📗 Math | Principal Component Analysis - theory and explanation
• ▶️ Demo | Principal Component Analysis - code

🤖 Non Negative Matrix Factorization

• 📗 Theory | Non-negative Matrix Factorization - theory and explanation

• ▶️ Demo | Non-negative Matrix Factorization - OOP's implement

• ▶️ Demo | Non-negative Matrix Factorization - simple implement from scratch

🤖 Singular Value Decomposition

• ▶️ Demo | SVD - from scratch

Clustering

Clustering is the task of dividing the population or data points into a number of groups such that data points in the same groups are more similar to other data points in the same group and dissimilar to the data points in other groups.

🤖 K-Mean

• 📗 Theory | K-Mean - theory

• ▶️ Demo | K-Mean - from scratch

🤖 DBSCAN

• ▶️ Demo | K-Mean

Deep Learning

🤖 Perceptron

Perceptron is similar to SVM it also construct a hyper plane in high dimension if data is linearly seperable.

• 📗 Theory | Perceptron - Theory of perceptron
• ▶️ Demo | Perceptron - code

🤖 Artificial Neural Network

• 📗 Theory | Artificial Neural Network - Theory of ANN
• 📗 Math | Regression with Artificial Neural Network - theory and code
• ▶️ Demo | Bivariate Classification with Artificial Neural Network
• ▶️ Demo | Multivariate Classification with Artificial Neural Network using Sigmoid function
• ▶️ Demo | Multivariate Classification with Artificial Neural Network using ReLu function
• ▶️ Demo | Multivariate Classification with Artificial Neural Network ( DevnagariHandwrittenDataSet )- complete code from scratch ( Accuracy 73 % )
• ▶️ Demo | Multivariate Classification with Artificial Neural Network ( MNIST )- complete code from scratch (Accuracy 92 %)
• ▶️ Demo | Bunch of Activations

🤖 Convolutional Neural Network

• ▶️ Demo | CNN

🤖 Recurrent Neural Network

A recurrent neural network (RNN) is a class of artificial neural networks where connections between nodes form a directed graph along a temporal sequence. This allows it to exhibit temporal dynamic behavior.

• 📗 Theory | RNN - theory and explanation

• ▶️ Demo | RNN - Vanilla RNN for Single-Batch from scratch

• ▶️ Demo | RNN - Vanilla RNN for Multi-Batch from scratch

• 📗 Theory | RNN - Derivation of Back Propagation through Time(BPTT).

Optimization Algorithms

🤖 Gradient Decent

• 📗 Math | Gradient Decent
• ▶️ Demo | Multivariate Gradient decent

🤖 Gradient Decent Check

• ▶️ Demo | Gradient Decent Check

🤖 Gradient Decent with Mini Batch

• ▶️ Demo | Gradient Decent with Mini-Batch (Devnagri)

🤖 Gradient Decent with Adam Optimization

• ▶️ Demo | Gradient Decent with Adam optimization (MNIST)

🤖 Gradient Decent with Momentum Optimization

• ▶️ Demo | Gradient Decent with Momentum optimization

🤖 Gradient Decent with RMSProp Optimization

• ▶️ Demo | Gradient Decent with RMSProp optimization

🤖 Newton's Raphson Method

• 📗 Math | Newton's Raphson Method - Theory
• ▶️ Demo | Newton's Raphson Method - Implementation with ROC and AUC Curve

Paper Implement

🤖 Paper Implement

• LeNet
• AlexNet
• GoogLeNet
• ResNet

Deep Learning with Tensorflow

• ▶️ Demo | Deep Learning with Tensorflow

Regression

• ▶️ Demo

Classification

• ▶️ Demo

Complex Modelling using Functional API

• ▶️ Demo

Tensorboard

• ▶️ Demo

Hyperparameter Fine Tuning

• ▶️ Demo

Tensor and Operations

• ▶️ Demo

Custom Model Building

• ▶️ Demo

Loading and Preprocessing Large Data

• ▶️ Demo

CNN with Tensorflow

• ▶️ Demo

Sequential Modelling

• ▶️ Demo

Character Level Modelling

• ▶️ Demo

Stateless RNN

• ▶️ Demo

Stateful RNN

• ▶️ Demo

Word Level Modelling

• ▶️ Demo

Sentiment Analysis

• ▶️ Demo

Encoder-Decoder

• ▶️ Demo

BiDirectional Layer

• ▶️ Demo

Beam Search

• ▶️ Demo

Attention

• ▶️ Demo

Transformers Multi Head Attention

• ▶️ Demo

NLP with HuggingFace and Transformers

• ▶️ Demo

UnderComplete Linear AutoEncoder

• ▶️ Demo

Stacked AutoEncoder

• ▶️ Demo

Convolutional AutoEncoder

• ▶️ Demo

Recurrent AutoEncoder

• ▶️ Demo

Denoising AutoEncoder

• ▶️ Demo

Sparse AutoEncoder

• ▶️ Demo

Variational AutoEncoder

• ▶️ Demo

Generative Adversarial Networks

• ▶️ Demo

Deep Convolutional GAN

• ▶️ Demo

Hasing using Binary AutoEncoder

• ▶️ Demo

Denoising AutoEncoder 3 Channel Image

• ▶️ Demo

Model Deployment

• ▶️ Demo

Topic Modelling

• ▶️ Demo

Prerequisites

Installing Python

Make sure that you have Python installed on your machine.

You might want to use venv standard Python library to create virtual environments and have Python, pip and all dependent packages to be installed and served from the local project directory to avoid messing with system-wide packages and their versions.

Installing Dependencies

Install all dependencies that are required for the project by running:

Datasets

The list of datasets that are being used for Jupyter Notebook demos may be found in DataSet Folder.

Clone

git clone https://github.com/Girrajjangid/Machine-Learning-from-Scratch.git