Original article was published on Artificial Intelligence on Medium
Machine Learning is one of the most trending topics in nowadays technology and involves many aspects of our lives, so talking about it in daily conversations is getting more common. But it is also clear that having a talk like that with grandparents may not be the easiest, because they aren’t quite familiar with tech these days, especially because of the fast growth of development in this field which supposes a thousand new terms and devices that would be difficult to know all at once. So, I will omit much of the technical lingo and make this explanation lighter to get the basics on this topic.
What is Machine Learning?
This concept is not easy to define but everyone will probably agree that it is an Artificial Intelligence (AI) subfield, so let’s see this concept first:
AI can be thought of as simulating the capacity for abstract, creative, deductive thought — and particularly the ability to learn — using the digital, binary logic of computers. — Bernard Marr
Machine learning focuses on the development of algorithms that can access data and use it to learn for themselves.
In simple terms, Machine learning is a subset of Artificial Intelligence (AI) which provides machines the ability to learn automatically & improve from experience without being explicitly programmed to do so. — Sisodia, Dilip Singh, Pachori, Ram Bilas, Garg, Lalit
If we think of the daily tasks we do, we can compare the learning process of machines with our learning process to choose vegetables and fruits when we visit the grocery store. It turns out that we didn’t know how to pick the best veggies the first time we went shopping, it took a few visits and analyzing the color, size, and texture among other characteristics, to make the best decision to buy. Similar to humans, Machine Learning allows machines to learn from previous experiences to solve problems by making smarter decisions later in the future.
Need For Machine Learning
A few reasons why Machine Learning is so important:
- Increase in Data Generation: Due to nowadays excessive production of data, Machine Learning comes in useful because can be used to structure, analyze, and draw valuable insights from lots of data.
- Improve Decision Making: By making use of several algorithms, Machine Learning can be used to make better business decisions, that is the case of forecast sales, predict downfalls in the stock market, etc.
- Identify patterns and trends in data: This is the most essential part of Machine Learning, by building predictive models and using statistical techniques, it allows humans to dig beneath the surface and explore the data at a minute scale unlike doing manually which will take days.
- Solve complex problems: From detecting the genes linked to the deadly ALS disease to building self-driving cars, Machine Learning can be used to solve the most complex problems.
Who’s using it?
- Financial services: Banks and other businesses in the financial industry use machine learning technology for two main purposes: to identify important insights in data, and prevent fraud.
- Government: Government agencies such as public safety and utilities have a particular need for machine learning since they have multiple sources of data that can be mined for insights. Analyzing sensor data, for example, identifies ways to increase efficiency and save money.
- Health care: Machine learning is utilized in wearable devices and sensors that can use data to assess a patient’s health in real-time. It can also help medical experts analyze data to identify trends or red flags that may lead to improved diagnoses and treatment.
- Retail: Websites recommending items you might like based on previous purchases are using machine learning to analyze your buying history to personalize a shopping experience, implement a marketing campaign, price optimization, merchandise supply planning, and customer insights.
- Oil and gas: Finding new energy sources. Analyzing minerals in the ground. Predicting refinery sensor failure. Streamlining oil distribution to make it more efficient and cost-effective.
- Transportation: Analyzing data to identify patterns and trends is key to the transportation industry, which relies on making routes more efficient and predicting potential problems to increase profitability.
Machine Learning Process
As in the image below, the Machine Learning process begins by feeding the machine a lot of data, by using this data it is trained to detect hidden trends and insights which are then used to build a Machine Learning Model by using an algorithm in order to solve a problem.
This process involves building a predictive model that can be used to find a solution to a problem statement. The next steps are followed in a Machine Learning process:
Step 1: Define the objective of the Problem Statement
Step 2: Data Gathering
Step 3: Data Preparation
Step 4: Exploratory Data Analysis
Step 5: Building a Machine Learning Model
Step 6: Model Evaluation & Optimization
Step 7: Predictions
How exactly do machines learn?
Machine Learning algorithms give machines the power to learn very complex models and to process a great quantity of information. On the other hand, they are entirely dependant and limited by the learning input (the raw data).
So in order to bring value, Machine Learning must be associated with other disciplines like Computer Science (Big Data) and domain expertise. It is the combination of these disciplines what is called DataScience.
In Machine Learning there are mainly three possible ways of learning things through data: Supervised Learning, Unsupervised Learning, and Reinforcement Learning.
To understand Supervised Learning let’s consider this analogy. When we were learning math as kids we all needed the guidance of our teachers to solve the problems. They helped us understand what math operations are and how to do them. In a similar way, we can think of supervised learning as a type of Machine Learning that involves a guide. In this case, there is a labeled data set that works as the teacher that will train the machine to understand patterns in the data and is nothing but the training data set.
Consider the above figure. Here we are feeding the machine with images of dogs and cats and the goal is for the machine to identify and classify the images into two groups (dog images and cat images). The training data set that is fed to the model is labeled, as in, we are telling the machine, ‘this is how a dog looks and this is a cat’. By doing it you are training the machine by using labeled data. In this type of learning, there is a well-defined training phase done with the help of the labeled data.
You can think of unsupervised learning as a very smart kid that learns without the need for any guidance. Unlike the Supervised Learning, the model is not fed with labeled data, as in the model has no clue that says ‘this image is a dog and this is a cat’, instead it figures out patterns and the differences between dogs and cats on its own by taking in great amounts of data.
Continuing with the above example, the machine identifies prominent features of the dog such as hairy ears, bigger size, etc, to understand that this image is of type a. Then as well, it finds the corresponding features in the cat and knows that this image is of type b. Thus, it classifies the images into two different classes without knowing which one is the dog or the cat.
This type of Machine Learning is relatively different than the previous two and its approach is a bit like the way a baby learns things. In fact, at the start, the machine does not know anything.
So how does it work? Take a look at the image above. In Reinforcement Learning, an ‘agent’ (the machine) will take decisions step by step thanks to the data it will get from its environment. Then the environment will answer a Reward (it will tell whether the action was a good one or a bad one). According to the reward, the agent will alter its rules of decision to act better in the future. The long term goal of Reinforcement Learning is to maximize rewards.