- To keep track of my work
- To prepare a document sequentially based on my learning, so that It will help to other newbie's.
Today's Progress :
-
Basic understanding of Machine learning - source: Andrew Ng course from Coursera
-
Types Machine learning
- Supervised learning - The main goal in supervised learning is to learn a model from labeled training data that allows us to make predictions about unseen or future data.
- Unsupervised learning - Here we deal with un-labelled data of unknown structure. This technique allows pile of information into meaningful subgroups (clusters) without having any prior knowledge of their group memberships.
- Reinforcement learning - In reinforcement learning, the goal is to develop a system (agent) that improves its performance based on interactions with the environment
-
Importing libraries
-
Reading datasets using pandas and numpy libraries
-
Read about Data Preprocessing
- Handling missing data in dataset
- Handling Categorical variables with OneHotEncoders Since Machine Learning models are based on mathematical equations, so it can cause problems with this kind of variables.
Today's Progress :
- Practiced handling missing data - Link for code
- Practiced handling categorical variables using OneHotEncoder from sklearn - Link for code
Today's Progress :
- Splitting dataset into train and test - Link for code
- Feature scaling
Today's Progress :
- Started reading about Linear Regression
Today's Progress :
- Continued reading about Linear Regression
Today's Progress :
- Completed reading about Linear algebra toipc - Andrew Ng course link
Today's Progress :
- Completed Week 1 from Andrew Ng course from Coursera - Week 1
Today's Progress :
- Read about Introduction to Computer vision
Today's Progress :
- Read about,
- Viola & Jones Algorithm
- Haar like features
- Integral Image
Today's Progress :
- Read about,
- Started Installing openCV from http://www.codebind.com/linux-tutorials/install-opencv-ubuntu-18-04-lts-python/, all steps are done, and still not able to import it. Will debug this tomorrow.
Today's Progress :
- Read about Training Classifier
- Read about Adaboost boosting algorithm
Today's Progress :
- Worked on Basic face and smile detection using opencv, But It is not very accurate. Here is the code
Note: Does anybody know, which parameter I should be changing to increase the accuracy.
Today's Progress :
- Read about SSD (Single Shot Multibox Detector) Algorithm and Predicting Object position.
Today's Progress :
- Read about Object Detection with SSD
Today's Progress :
- Read about Gradient descent Gradient descent is a optimization algorithm used to find the values of parameters (coefficients) of a function (f) that minimizes a cost function.
- Read about back propagation
Today's Progress :
- Read about Cross Validation -- We do Cross validation after training the model to test the error rate. If error rate is low it means we have builded a good model, If error rate is high then model is not good. -- This is one method where we can check the performance of the model -- It helps in selecting the best fit model -- It ensures that model is not over fit
- Types of Cross Validation
- Hold Out Method
- K-Fold CV
- Leave One out CV
- Bootstrap method
k - fold cross validation variance and bias trade off
- Started with Titanic survival prediction challenge in Kaggle
- Read about KNeighborsClassifier
- Continued working on Kaggle Titanic dataset Code
- Learned about productionizing Machine learning models with Flask, Gunicorn and Nginx
- Completed Titanic Kaggle challenge with 78% Model accuracy
- Read about Feature Engineering