MachineLearningFromScratch

This repository is a group of all machine learning/artificial intelligence algorithms I have experimented with. Each notebook is self-contained with relevant information.

Important Functions

1. [✅] Softmax
2. [✅] Sigmoid
3. [✅] MinMax Scaling
4. [✅] Euclidean Distance
5. [✅] Mahalanobis Distance
6. [✅] Cross-Entropy Loss
7. [✅] Gradient Descent
8. [] ReLU

Linear Regression:

1. [✅] Using NumPy
2. [✅] Using TensorFlow
3. [✅] Using Scikit learn
4. [✅] Using PyTorch
5. [✅] For Hypothesis testing

Logistic Regression:

1. [] Using Python
2. [✅] Using NumPy
3. [] Using TensorFlow
4. [] Using Sci-Kit Learn
5. [] Using PyTorch

Time Series Forecasting using ARIMA

1. [✅] ARIMA on TCS stock value

Decision Tree:

1. [] Using Python
2. [] Using NumPy
3. [] Using TensorFlow
4. [] Using Sci-Kit Learn
5. [] Using PyTorch
6. [] For classification

Random Forest

Boosting

1. [✅] Adaboost for Binary Classification
2. [] Extreme Gradient Boosting

Neural Networks

1. [] Using Python

Clustering

1. [] KMeans
2. [] Density Based
3. [] Hierarchical

Bayesian Network

1. [✅] Demonstration of a Bayesian Network

Reinforcement Learning

1. [✅] Model Based RL- Value Iteration, Policy Iteration
2. [] Model Free RL- Monte Carlo, Bootstrapping, Temporal Difference
3. [] Model Free RL- Q-learning
4. [] RL on Dataset

Diffusion Modelling- Link

1. [✅] Zero-Shot Classification using CLIP
2. [✅] Zero-Shot Classification on Distorted Image using CLIP
3. [✅] Text to Image using Stable Diffusion Pipeline
4. [✅] Custom Noise Schedule for Image-to-Text Generation
5. [✅] Interpolation of Latent Space to transition between two prompts
6. [✅] Manipulation of Latent Space to impact specific attributes of a generated image
7. [✅] Parameter tuning of guidance scale parameter
8. [✅] Guided Image Generation using ControlNet
9. [✅] Improving an image by refining gradients

Transformer Architecture

1. [] Next Word Prediction From Scratch
2. [✅] Next Word Prediction Using Finetuning on GPT2

Paper Implementations

1. [✅] Contrastive Language Image Pretraining (CLIP)
2. [] Deep Imbalanced Regression (DIR)
3. [] Generative Adversarial Networks
4. [] Attention is all you need