The Ultimate Guide to Understanding
Machine Learning (ML)

Introduction to Machine Learning (ML)

From the Terminator to HAL, science fiction is chalked full of futuristic machines capable of acting autonomously and of learning, often better than human beings. And as technology advances, these machines are becoming less science fiction—today, they’re simply science.

But fully autonomous and intelligent machines aren’t quite a reality yet. One of the major differences currently marking a divide between human and machine intelligence is a computer’s inability to think critically—at least, not quite in the way that we do. This line is being blurred, however, with recent advances in a program or a machine’s ability to adapt its own parameters of judgement to optimize solutions, or to put it more colloquially, a machine’s ability to learn. Appropriately, we call this Machine Learning.

Machine Learning (ML) Definition

What is the defintion of Machine Learning (ML)?

Simply put, Machine Learning is a field of computer science that focuses on getting machines to “learn” and to continually develop—autonomously.

Carnegie Mellon puts it this way:

“The field of Machine Learning seeks to answer the question ‘How can we build computer systems that automatically improve with experience, and what are the fundamental laws that govern all learning processes?’”

Would you like to learn more about commonly used words in Machine Learning Terminology? Visit our Machine Learning Glossary to discover must know industry terms. 

Machine Learning (ML) Terminology

What are some of the terminology of Machine Learning (ML)?

Deep Learning: This is a subset of machine learning (which is a subset of AI) where the neural networks are layered to support greater computing power—both to process data more efficiently and to process unstructured data.

Artificial Intelligence (AI): This refers to the field of study and the creation in computer science of a program or machine capable of replicating thought.

Decision Trees: These are intricate if/then models that branch out in a system’s programming, mapping out how the machine will make determinations. These trees continue to grow and change as the machine learns.

Neural Network: These are sequences of complex algorithms that generate and “map out” a machine’s functions—both for input and output data.

Artificial Neural Networks (ANN): These are inspired by the neural networks found in our brains, and are the means of “learning” by massive association (or forming connections and learning relationships without being explicitly told how to do so).

Classification: Each piece of data is seen by a learning machine as an observation; and the cataloguing of these observations is called classification.

Learning Rate: The learning rate measures how quickly a machine can abandon old configurations or decision trees for new ones. The old might be incorrect, or simply less effective than the new.

Natural Language Processing (NLP): This is a system or machine’s capacity to break down, then recognize, then assign meaning to, learn from and act on natural language as delivered by users.

Clustering: Before observations can be classified by the machine, first they have to be grouped into categories and sub-categories through intuitive clustering.

Machine Learning (ML) Explained

What is Machine Learning (ML)?

The term “Machine Learning” was coined in 1959 by Arthur Samuel, a prominent figure in the origins of Artificial Intelligence when it became a formal field of research. And as the field began to develop, it drifted from the more symbolic and “fantastical” ideas of Artificial Intelligence and edged closer to more practical methods driven by the Statistics and Probability Theory. Artificial Intelligence not only established itself as its own field of research, but here it developed many subfields like Deep Learning and Computer Vision.

Machine Learning is similar to Data Mining, and the two are sometimes confused. The difference comes in that Data Mining is the practice of discovering unknown properties embedded in certain sets of data, while Machine Learning is a means of producing predictions based on already known properties of data sets.

Additionally, Machine Learning has very strong ties to Optimization. In fact, the goal of Machine Learning is often to minimize a certain loss function, or the discrepancy between a prediction and the actual result. Put simply, Machine Learning attempts to optimize the way its very predictions are made.

There are many different approaches to Machine Learning. Among the most popular today are:

Decision Tree Learning

In the context of Machine Learning, these decision “trees” are a set of points of information referred to as “nodes,” connected by edges referred to as “branches.” A decision tree, then, is a tree that outlines criteria in each node that lead the machine to the best option found in the bottom node of the path taken—this bottom node is referred to as a “leaf.”

A certain decision tree can be “learned” by a machine or program by taking the set of decision criteria and splitting the set into subsets at each point along the path, thereby partitioning the data. Once this tree has been formed, the decision process is simple—the machine simply follows a path down the tree based on the relevant criterion.

Artificial Neural Networks

Artificial Neural Networks are inspired by the biological Neural Networks found in our brains, and are the essential means of “learning” by massive association (or forming connections and learning relationships without being explicitly told how to do so).

An easy example of this is image recognition. If you have a smartphone, chances are you’re able to search for certain objects, and your phone will show you the images you have with this object. This ability can be learned, for example, by giving a program loads of images with apples and without apples, and then manually categorizing them as such. The program analyzes all the images and sees what they have in common and how they differ, building connections between objects that are apples and objects that aren’t, eventually learning to tell apples from oranges.

Deep Learning

While Deep Learning can take other forms, the most common form it takes is Artificial Neural Networks, or “layered” learning. The “layered” refers to how many times the data is analyzed and transformed into information that the machine or system can use. In image processing, for example, these layers can mean breaking an image down into pixels, then shapes, then recognizable objects, etc.

Support Vector Machines

Support Vector Machines, also known as SVMs, are fed data on examples of items that fit into certain categories. This acts as machine “training,” and teach the systems to differentiate between these categories. This method is useful in image and text categorization.

Clustering

Cluster Analysis, or Clustering, groups data together into different “clusters” based on similarities and differences found between the data. These clusters are then analyzed to “learn” more about how the elements of the data within clusters are related and how they aren’t. This method is useful in genetic research as well as crime and climate analysis.

Machine Learning (ML) News

Latest developments in Machine Learning (ML) news

Machine Learning (ML) Categories

What are the categories of Machine Learning (ML)?

After seeing some of the different approaches Machine Learning can take, it should be easy to see that there are a several different kinds of Machine Learning. Typically, Machine Learning can be broken up into four different categories:

• Supervised Learning
• Unsupervised Learning
• Semi-supervised Learning
• and Reinforcement Learning

Category A: Supervised Learning

Supervised Learning is based on human intervention. A user feeds the program data and then judges how “good” the outputs are. Telling the program which outputs are better than others teaches the program what and how to judge, or what elements should be given more or less weight. Once the algorithm has been trained or “taught,” it can take its honed judgement and apply it to different data sets. Neural Networks and Decision Trees are examples of Supervised Learning.

Category B: Unsupervised Learning

Unsupervised Learning, as one might expect, does not involve much (or any) human intervention. When an algorithm learns “unsupervised,” it’s given loads of manually-labeled examples and draws its own parameters from the observed data. Clustering algorithms are an example of Unsupervised Learning.

Category C: Semi-supervised Learning

Semi-supervised Learning is basically a combination between Supervised and Unsupervised Learning. It is practiced when the data set contains some valuable information regarding the grouping parameters, but not all of the input data is labeled. These situations have become more common, as having a fully-labeled input set requires somebody to label all of the data, and in many cases this can be too time-consuming or expensive.

Category D: Reinforcement Learning

Reinforcement Learning is a familiar method for most of us, as it mirrors the process of training a pet and other like learning. This type of Machine Learning is applied when the problem at hand is, in and of itself, the task of making the correct decision. The program will decide on an action to take, the environment will then either reward or disincentivize the decision, and the program will adjust its parameters accordingly. The decision is based only on what the program knows in that moment, which is cumulative in nature but without memory. This is a Markovian process.

This type of learning can be observed in online and computer board games where the program learns what moves are “good” and “bad” based on how it’s doing in the game. For example, if the computer is playing chess and a certain move leads to its queen being captured, the program will learn that such a move isn’t a good one. In future iterations, the move will not be repeated.

How Does Machine Learning (ML) Work?

How Machine Learning (ML) Works?

Machine Learning systems are comprised of three major and crucial components:

• Model
• Parameters
• Learner

The model of the Machine Learning system is that which makes predictions. Whether it be through trees, clusters, or neural networks, the model is the manner in which the decisions are made.

The parameters answer why the decisions are made. These can be contained in the nodes of decision trees or simply as conditions listed out for comparison. We can think of the parameters as the conscience, or as the values that drive every decision taken.

The learner in a Machine Learning system is the way a program adjusts its parameters for future decisions and predictions. In a supervised environment, this can be through human involvement. And in an unsupervised environment, this can be by automatic adjustment through a predetermined optimization method.

Machine Learning (ML) Key Components

Components of Machine Learning (ML)

Data

Data is simply information, and is an absolutely crucial part of Machine Learning. In the end, Machine Learning is nothing more than improving predictions based on studied data.

Depending on the subject at hand, the amount of data available can be astonishing. With the help of devices in our everyday lives, we’ve managed to create more data in the past 10 years than in all of humanity before then. Some of these data sets are so large that they entirely changed our concept of data analysis and our approaches to study it, birthing new fields of study (like Machine Learning) into the picture. These seemingly-fathomless data sets are known as Big Data.

Parameters

As mentioned before, parameters are a the program’s way of making decisions. Think of a baby trying spicy food for the first time. Maybe the babe didn’t even hesitate putting the food in his mouth, but as soon as the spice hit, something in his decision process was rewired to think twice before making that decision again.

Whether it’s from reinforcement and reward or from loss-minimization optimization, a program will adjust its parameters similarly, “learning” to make decisions that lead to better outcomes.

Users

Of course, people are necessary for the concept of Machine Learning to exist, but more than that, they play an important role in the process for many machines. In a supervised learning system, humans are needed to provide feedback and help the system learn. In an unsupervised learning system, humans are still needed to categorize and label the initial set of training data.

Problems

If you’re looking for a better solution, then there must be a problem you’re trying to solve. Likewise, in order for a Machine Learning system to exist, there must be some sort of problem that needs to be solved. This can range from wanting to answer simple questions all the way to mapping out genomes. In short, for every answer, there was first a question that inspired the problem-solving process.

Machine Learning (ML) Companies

Discover innovative Machine Learning (ML)startups and companies

It takes bold visionaries and risk-takers to build future technologies into realities. In the field of machine learning, there are many companies across the globe working on this mission. Our mega list of artificial intelligence, chatbot, computer vision, natural language processing, and machine learning companies, covers the top companies and startups who are innovating in this space.

Machine Learning (ML) Applications

What are applications of Machine Learning (ML)?

Machine Learning is a means through which predictions can be increasingly accurately made based on often enormous sums of data. Perhaps that sounds vague, but truth be told, the applications of this science are broader today than ever. Machine Learning has applications in many, many fields and studies. Here are just a few:

Search Engines

You’ve probably noticed that your favorite search engine tends to recommend things that might interest you. This isn’t your phone listening in on your conversations—in fact, it’s a much more complicated process. Each search engine collects data from your searches and your browsing history, compares the data points, and draws predictions as to what you like to see, what you don’t like to see, and what you’re likely to need and want. If you’ve ever felt shocked or surprised at the accuracy of a suggestion on your search engine, you can take it as a testament to the efficiency and accuracy of today’s Machine Learning.

Economics

Economic analytics focuses around trying to see what has happened or what will happen to certain variables, that in turn depend on certain variables, that also depend on certain variables that depend on certain variables…It’s an enormously complicated and often unpredictable system, especially on a global scale, and taking in the massive data available to try and parse out the millions of subtle relationships is a task best suited for machine learning.

Machine Learning is becoming more and more popular in economics today. While experts say that it hasn’t yet completely revolutionize the field, its worth in ongoing research is nothing to be scoffed at.

Marketing

Similar to search engines, many aspects of marketing employ Machine Learning to predict what each consumer is most likely to be interested in. This can be anything from emailed coupons to targeted advertisements. Again, if you’ve ever seen an ad on your phone that is hauntingly specific to your last search or conversation, it probably was.

Machine Learning has been a powerful tool in marketing for a while now, such as with the growing-in-popularity chatbots. Another infamous example can be found with Target’s coupon system; several years ago, Target realized a customer was pregnant before her own father had.

Computer Vision

Have you ever thought about how your vision works? It’s not something we typically think about seeing as it’s such a natural part of daily life, but the process far from trivial. Recreating vision is difficult, and recreating the ability to interpret and almost immediately react to that interpretation is another level of difficulty altogether. But through processes like image processing and recognition, Machine Learning (Deep Learning, specifically) has inspired and allowed sizeable advances in this field.

Machine Learning (ML) Tools

What are Machine Learning (ML) tools?

Shogun

This is considered by many to be the oldest Machine Learning tool, and has undergone massive makeovers with time. Shogun is written in C++ but it can be used in Python, Java, Octane, Ruby, R, and Matlab. Shogun offers tools for regression, visualization, model-selection strategies, pre-processing, one-time and multi-class classification.

Apache Mahout

Apache Mahout is an open source and completely free resource. Created on Apache Hadoop, Apache Mahout has tools that are useful for recognizing certain patterns in large data sets, all in an accessible R syntax.

Scikit-Learn

Built on top of Python packages, Scikit-Learn can help you with both data analysis and data mining. It is open source and available under a BSD License, developed by Machine Learning experts.