Skip to content

barathvaj/Machine-Learning-From-Scratch-CodingChallenge

Repository files navigation

Machine Learning-CodingChallenge

I am conducting a coding challenge session , objective here is to solve problems using Machine Learning from Scratch(just using numpy).

Session 1 : Linear Regression

About : In this part of this exercise, you will implement linear regression with gradient desent to predict the brain weight based on Gender, Age and head size.

Link of Work: Commit

References: Commit

Session 2 - Logistic Regression

About : In this exercise, we will implement logistic regression and apply it to two different datasets. One in which data is linear and other in which data is non linear. Here we also see regularization and apply polynomial features to handle non linear dataset

Link of Work: Code

References: Code

Session 3 - Logistic Regression For Multi-Class Classification

About : In this execise, we will implement logistic regression for Multi-class classification influenced by concept provided by ndrew ng on one-vs-all. Here we will recognize handwritten digits(from 0 - 9). Here we will extend our previous implemen- tion of logistic regression and apply it to one-vs-all classification.

Link of Work: Code

Session 4 - Neural Networks From Scratch

About : Objective here is to simple neural network using numpy to predict with simple Toy Dataset and create a object model like CNTK, Pytorch so that neural network can created with ease.

Link of Work: Code

Session 5 - Neural Networks From Scratch To Recognize Handwritten Number

About : It is an extention to the neural network which we built in last session and this includes creation of multiple activation/loss functions, regularization and create a model to predict handwritten digits(0-9)

Link of Work: Code

References: Code

Session 6 - With DropOut - Neural Networks From Scratch

About : In this session, we will discuss dropout in detail and compare it with ensemble or bagging techniques. Extend our neural network which we builded before to support Dropout.

Link of Work: Code