Linear regression may be both the simplest and most popular among the standard tools to regression.Dating back to the dawn of the 19th century, linear regression flows from a few simpleassumptions. First, we assume that the relationship between the independent variables x and thedependent variable y is linear, i.e., that y can be expressed as … Continue reading Basic Elements of Linear Regression Model

## Affine Function

Affine functions is a vector-valued functions of the form f(x_1,…,x_n)=A_1x_1+…+A_nx_n+b These coefficients can be scalar or dense or sparse matrices. The constant term is a scalar or a column vector. In geometry, an affine transformation or affine map (from the Latin, affinis, "connected with") between two vector spaces consists of a linear transformation followed by … Continue reading Affine Function

## Write with transformer

When I asked a GPT-2 transformer about "what is life", We can see how beautifully this model portray its thought about life. This is my first write up from GPT-2. Will post more such stories using these transformer models. "Life is suffering and pleasure. You are suffering for the pain you have caused yourself. Your … Continue reading Write with transformer

## Vector Norm in Machine Learning

Some of the most useful operators in linear algebra are norms. Informally, the norm of a vectortells us how big a vector is. The notion of size under consideration here concerns not dimensionalitybut rather the magnitude of the components. In linear algebra, a vector norm is a function f that maps a vector to a … Continue reading Vector Norm in Machine Learning

## Softmax Operation

The main approach that we are going to take here is to interpret the outputs of our model as probabilities. We will optimize our parameters to produce probabilities that maximize the likelihood ofthe observed data. Then, to generate predictions, we will set a threshold, for example, choosingthe label with the maximum predicted probabilities. Put formally, … Continue reading Softmax Operation

## A Super Harsh Guide to Machine Learning:

Found this discussion on reddit, its sure helpful for data science roadmap. https://www.reddit.com/r/MachineLearning/comments/5z8110/d_a_super_harsh_guide_to_machine_learning/ Keep Learning

## CODE A NEURAL NETWORK IN PLAIN NUMPY Part 2: Planar data classification with one hidden layer

In the last post we have seen neural network with only two layers that is "Input layer" and "Output layer", which is like a logistic regression algorithm. However in this post we are going to code a Neural network with one more layer that is "hidden layer". You will learn how to: Implement a 2-class … Continue reading CODE A NEURAL NETWORK IN PLAIN NUMPY Part 2: Planar data classification with one hidden layer

## Code a Neural Network in plain NumPy: Part 1 (with no Hidden Layer)

You will learn to: Build the general architecture of a learning algorithm, including: Initializing parameters Calculating the cost function and its gradient Using an optimization algorithm (gradient descent) Gather all three functions above into a main model function, in the right order. Overview of the Problem set Problem Statement: You are given a dataset containing: … Continue reading Code a Neural Network in plain NumPy: Part 1 (with no Hidden Layer)

## Early Stopping in Neural Networks

Early Stopping is to stop the Training of Neural Networks at the Right Time or Stop training when a monitored quantity has stopped improving. A major concern with training neural networks is in the choice of the number of training epochs to use. Too many epochs can lead to overfitting of the training dataset, whereas … Continue reading Early Stopping in Neural Networks