Course Description
This is an in depth hands on workshop based course that teaches the fundamentals of Machine Learning (ML) and Neural Networks (NN). The course is designed for professionals who will be asked to solve various business problems using ML. Workshop projects are designed to be realistic. They follow a complete lifecycle:
- Collect and clean up data.
- Design a model
- Train the model with data
- Start doing prediction
We primarily use Keras for the workshops. In some cases, we dive lower and code using Tensorflow API. This is done to gain a better understanding of the concepts.
Course Outline
Workshop 1 - Tensorflow Basics
Learn about computation graph, variable, placeholder and matrix.
Workshop 2 - Simple Linear Regression
Perform linear regression in a very simple problem domain. The goal is to learn how linear regression works.
Workshop 3 - AirBnB Property Price Prediction
This is a realistic regression problem. We try to predict property rental prices in the Boston area. We learn to work with categorical features like neighborhood and property type.
Workshop 4 - Build a Simple Neural Network
We will learn to build a basic neural network to solve a very simple problem. Again, the goal here is to understand the fundamentals of NN.
Workshop 5 - AirBnB Property Price Prediction Using Neural Network
We now use NN to solve this problem with higher accuracy.
Workshop 6 - Basic Binary Linear Logistic Regression
We learn to do binary classification using linear logistic regression.
Workshop 7 - Titanic Survivability Prediction
This is a more realistic binary classification problem. This also uses categorical features like class of travel (first class, second class etc.).
Workshop 8 - Multi-class Linear Logistic Regression
We learn to classify among multiple classes. This workshop uses fetal heart monitoring (Cardiotocography) to predict complications during childbirth.
Workshop 9 - Multi-class Logistic Regression Using neural Network
We solve the same fetal heart monitoring problem using neural network. This gives us much higher accuracy.
Workshop 9 - Basic Convolutional Neural Network (CNN)
The goal of this workshop is the understand the structure of a CNN. We learn about the convolution layer, max pooling layer, fully connected layer and readout layer. We solve the MNIST handwritten digit comprehension problem.
Workshop 10 - See Convolution in Action
We go deeper into CNN. We actually observe how convolution works.
Workshop 11 - Solve CIFAR-10 Challenge
CIFAR-10 dataset consists of 60000 32x32 colour images in 10 classes. We train a CNN that tries to classify images in those 10 classes.
Duration
2 Days | 4 NightsApplies Towards the Following Certificates
- Data Science Certificate : Data Science
*Academic Unit eligibility to be determined by college/university in which you are enrolled in a degree seeking program.