Lectures / demos  

Lectures

Linear Regression Intuition

Linear regression is used to predict real-valued output as opposed to classification where data is predicted in predefined number of classes.  

Gradient Descent Algorithm 

Gradient descent is a first-order iterative optimization algorithm for finding a local minimum (extremum) of a differentiable function.  

Linear Regression using Gradient Descent

Topics covered in the lecture:

Intuition of linear regression, Formalizing cost function, discussion on gradient descent optimization algorithm and applying gradient descent to optimize parameters for linear regression. 

K-Nearest Neighbors algorithm

K-Nearest Neighbors algorithm (or KNN) is one of the most used learning algorithms due to its simplicity. It finds non-linear surface to separate data points from different classes.

K-Nearest Neighbors algorithm and Curse of Dimensionality

K-Nearest Neighbors algorithm (or KNN) is one of the most used learning algorithms due to its simplicity. It finds non-linear surface to separate data points from different classes. Curse of Dimensionality refers to a set of problems that arise when working with high-dimensional data.  

Perceptron and Perceptron Learning Algorithm

The perceptron was intended to be a machine, rather than a program. In a 1958 press conference Rosenblatt (inventor of Perceptron) made statements about the perceptron that caused a heated controversy. He proposed that perceptron to be the embryo of an electronic computer that expects will be able to walk, talk, see, write, reproduce itself and be conscious of its existence.

Machine Learning: Support Vector Machines (SVM) - A brief introduction


Support vector machines (SVMs) are perhaps the most used pre-deep learning, machine learning method. This video introduces basic concept of Support Vector Machines (SVM). Goal of SVM is to maximize distance between classes while classifying all training examples correctly. 

What is Machine Learning?


In this video I tried to explain three W's of Machine Learning: 1. What is Machine Learning 2. Why Machine Learning is useful? 3. How Machine Learning is different from other programming paradigms? 

Machine Learning - Model Evaluation


This lecture is recorded from online class and is  partially in English and partially in Urdu / Hindi, but hopefully will deliver concept to general audience.


Topics Covered: 

1. Data partitioning techniques

2. Measure of model evaluation 

DEMO

Machine learning / AI based system that detects whether a person is wearing a mask or not?

 The world Health Organization (WHO) and the governments covering 90% of the world’s population recommends that everyone wears a mask over their mouths and noses when venturing out of their homes into places where it is difficult to maintain distance from other people as it reduces risk of Coronavirus spread. 

Real-time eye blink detection



This demo is created using paper "Real-Time Eye Blink Detection using Facial Landmarks" by Tereza Soukupova and Jan Cech. 

Magic wand: fun drawing application for kids


OpenCV based real-time color tracking: fun drawing application for kids 

Real-time facial landmark points detection and tracking


Technologies used: 1. Python 2. Dlib 3. openCV 

Real time facial expressions recognition


Generally, facial expression recognition (FER) is based on extracting features from the face. Computer vision research community is generally lacking to infuse the information on how human visual system (HVS) does the same job effortlessly and in real-time. In my research I have proposed algorithms for FER which are based on HVS. 

Facial Expression Recognition using Pyramid of LBP



Demo video for the paper "Framework for reliable, real-time facial expression recognition for low resolution images". Paper is published in the journal of "Pattern Recognition Letters" Volume 34, Issue 10, 15 July 2013, Pages 1159-1168.