Advanced Machine Learning with ENCOG - Part 2
Course info
Course info
Description
Finding patterns in a multidimensional dataset has always been a challenging task, but self-organizing maps can simplify this process and can help to find interesting patterns and inferences. In this course, you will learn not only the fundamentals of self-organizing maps but also the implementation in a C# application using the ENCOG machine learning framework. In this course, you will also learn to use Hopfield networks in a pattern recall and reconstruction application. This course will also provide a real world case study on time series forecasting, where you will learn to forecast future behavior using historical values. The course also covers another very important aspect of machine learning: optimization. You will learn to solve optimization problems with the help of genetic algorithms. The concepts learned in this course are applicable for developers working in any other framework in any other language.
About the author
Abhishek Kumar is a data science consultant, author and speaker.He holds Master's degree from University of California, Berkeley.His focus area is machine learning & deep learning at scale.
More from the author
Section Introduction Transcripts
Unsupervised Competitive Networks
Hi, this is Abhishek Kumar, and welcome to the second module of the course, which is on Unsupervised Competitive Networks. Until now, we have been dealing with supervised networks where a training dataset usually had both input and ideal output parts, but many a times we do not have the luxury of ideal output values or class labels in the available dataset. But then also we would like to find some pattern among the dataset to get some inherent knowledge. In those scenarios, these unsupervised networks and methodologies will be very handy. This module will focus on first off the broad categories of unsupervised networks, which is competitive networks. By the end of this module, you will not only have a fundamental understanding of unsupervised learning and unsupervised networks, but also you will learn to get some interesting patterns inside a given training dataset with the help of ENCOG machine learning framework in the C# language.
Unsupervised Auto-Associative Networks
Hi, this is Abhishek Kumar, and welcome to the third module of the course, which is on Unsupervised Auto- Associative Networks. In the previous module, we covered the unsupervised competitive network, which was the first main category of unsupervised networks. In this module, we'll be taking another main category of unsupervised networks, which is auto-associative networks, which can deal with associative memories. Our brain functions in the same manner by recalling someone or something from our memory. By the end of this module, you will learn about the fundamentals of associative memory and its significance. You will also learn about one of the auto-associative networks, which is the Hopfield network, and how to implement the Hopfield network using the ENCOG machine learning framework. We will also take a C# demo at the end to further clarify the concepts.
Case Study: Time Series Forecasting
Hi, this is Abhishek Kumar, and welcome to the fourth module of the course. In this module, we will take a case study, which is on Time Series Forecasting, which means to predict future behavior by utilizing historical data values. Time series forecasting is arguably one of the most important applications on machine learning. In business scenarios, we refer to time as money because even a small amount of extra time for preparation can give you a competitive edge, and by predicting the future behavior about their sales or expenses, or other _____, companies can take various short term and long term _____ to leverage this competitive edge to increase their profits. By the end of this module, you will learn about some core concepts of time series forecasting and to explore the power of machine learning to make future predictions. You will also learn to utilize the ENCOG machine learning framework to build applications to predict future behavior.
Optimization Using Genetic Algorithm
Hi, this is Abhishek Kumar, and this is the fifth module of the course where we will take another aspect of machine learning that is optimization. In this module, you will learn another flavor of supervised learning, that is non-propagation supervised learning. So far, in various propagation supervised learning, we were having both input and ideal output values in the given training dataset, which we used to train the neural network. So in propagation supervised learning, the ideal output values supervise the whole training process, but in the non-propagation supervised learning, we will learn about another alternate mechanism to supervise the process, and thus eliminating the need of ideal output values. These non-propagation supervised learning techniques are widely used in optimization problems where we'll look for an optimized solution to any problem. In this module, you will learn about the fundamentals of non- propagation supervised learning and optimization problems. You will also learn a widely used optimization technique, that is genetic algorithm in this module. We will also take a C# demo to further understand the concepts and implement an optimization application using genetic algorithms in a real world scenario using the ENCOG machine learning framework.