Different Types of Machine Learning with Examples
Different Types of Machine Learning with Examples. Primarily speaking, machine learning is otherwise known as self learning machines or learning systems. It is an area of artificial intelligence devoted to algorithms that improve automatically through experience, or exposure to data. In addition, machine learning has it that it is divided into two basic areas such as supervised and unsupervised. Each has a specific purpose and activity, which will be presented in this article.
In this blog, we cover the main concept of what machine learning is. Furthermore, I discuss how machine learning works, and what major types of them are. I invite you to read the rest of the article about Different Types of Machine Learning with Examples.
What is Machine Learning ?
Source Image: iberdrola
Truly , the machine learning is a small application of artificial intelligence that allows machines to automatically learn and improve tasks, making them more productive. Based on the data collected, the machine refines the computer program to tailor it to the desired result. Because these machines are self learning, no explicit programming is required for these computers.
However, this concept has broadly and already entered our lives without us knowing it. Almost every machine we use, and the most advanced machines we’ve seen in the last decade, incorporate machine learning to improve product quality. Some examples of machine learning are self driving cars, advanced web search, and speech recognition.
How Does Machine Learning Work?
Next question is how does machine learning really work? Well, supervised learning, which has the ability to train the model on known inputs and outputs, is an interesting application. In addition, it’s made in such way, so that it predicts future results. While the other part is called unsupervised learning. Simple put, its job is to look for hidden patterns or internal structures in the input data. Look at the example below of how it works below:
Source Image: mathworks
Well, the above model, shows how ML, allows you to work with different types of algorithms and methods. These algorithms are created using various programming languages. Typically, a training data set is provided to an algorithm to build a model.
Now, given an input to the ML algorithm, it returns an estimated/predicted value based on the model. Now, if the prediction is correct, it is accepted and the algorithm is implemented. However, if the predictions are not accurate, we retrain the algorithm with the training data set to get accurate predictions/estimations.
Types of Machine Learning
Source Image: analytixlabs
Significantly, we use two types of learning methods. One is supervised learning, which trains a model on known inputs and outputs, so it predicts future outcomes, and the other is unsupervised learning. This method of learning looks for hidden patterns or internal structures in inputs.
Supervised Learning
Another method is supervised learning. This type has the interesting ability to pass historical inputs and outputs to the learning algorithm and the processing between each input/output pair allows the algorithm to manipulate the model that produces an output that is as close to the desired result as possible. Well, the common algorithms used in supervised learning include neural networks, decision trees, linear regression,and support vector machines.
Secondly, this type of machine learning gets its name because the machine is “controlled” during training. In turn, this means providing algorithmic information to aid learning. The output we feed to the machine is the labelled as data, and the rest of the information we provide is used as input features.
For example, if you’re trying to figure out the relationship between loan defaults and borrower information, you could give the machine 500 customer instances that defaulted on loans and another 500 customer instances that didn’t. Then, the labelled data “watches” your computer to get the information you’re looking for. Simply put, it is amazing and mind blowing technology advancement.
Use cases
Some examples of use cases include:
- Classifying whether banking transactions are fraudulent or not.
- Determining whether loan applicants are low or high risk.
- Predicting the failure of mechanical parts of industrial equipment.
- Finding disease risk factors.
- Predicting real estate prices.
Unsupervised learning
On the other hand we have supervised learning. In this instance, it requires the user to assist machine learning, while unsupervised learning does not use the same labelled training set and data. Instead, the machine looks for less obvious patterns in the data. For this purpose, this type is very useful when you need to uncover patterns and use the data to make decisions. Common algorithms used in unsupervised learning include hidden marked models, k-means, hierarchical clustering, and mixed Gaussian models.
Additionally, with this type of learning, it is widely used to build predictive models. Common applications also include clustering, which creates models that group objects based on certain properties, and association, which defines the rules that exist between clusters. Here are some usage examples:
Use cases
Reinforcement learning
The third type of ML is reinforcement learning. Here, with this type of machine learning, it is most similar to how humans learn. In brief, the algorithm or agent uses this type of learning by interacting with the environment and receiving positive or negative rewards. Here, the common algorithms include time delay, deep adversarial networks, and Q-learning.
Interestingly though, it is assumed, that most machine learning platforms lack reinforcement learning capabilities because they require more processing power than most organizations. Hence, the reinforcement learning is a domain that is ideally modelled and applied to static domains or to domains containing large amounts of relevant data.
Significantly, this type of machine learning requires less administration than supervised learning, and thus unlabelled datasets are considered easier to work with. Simultaneously, the practical applications of this type of machine learning are still emerging.
Use cases
Some examples of uses include:
- A robot is trained to learn policy using raw video images as input and used to replicate the actions that the robot saw.
- Dynamically controlling traffic lights to reduce traffic jams.
- Teaching cars to park themselves and drive autonomously.
Examples of Machine Learning
Source Image: bi-insider
Statistical arbitrage
In essence, statistical arbitrage is an automated trading strategy used in finance to manage large volumes of securities. This strategy uses trading algorithms to analyze a set of securities using economic variables and correlations.
Real world examples of statistical arbitrage:
- Identify real time arbitrage opportunities.
- Algorithmic trading which analyses a market microstructure.
- Analyse large data sets.
Extraction
Subsequently, another example of ML is extraction. In other words, machine learning extracts structured information from unstructured data. Organizations collect a huge amount of data from their customers. In particular, machine learning algorithms automate the process of annotating datasets for predictive analytics tools.
Real examples of extraction:
- Help physicians diagnose and treat problems quickly.
- Develop methods to prevent, diagnose, and treat the disorders.
- Generate a model to predict vocal cord disorders.
Predictive analytics
In the same way, machine learning classifies available data into groups. Then, groups are defined by rules set by analysts. After the classification is completed, the analyst calculates the probability of failure.
Examples of predictive analytics:
- Improve prediction systems to calculate the possibility of fault.
- Predicting whether a transaction is fraudulent or legitimate.
Speech recognition
Convert speech to text. Some software applications can convert live and recorded speech into text files. Speech can also be segmented by the intensity of time frequency bands.
Real world examples of speech recognition:
- Voice search.
- Voice dialling.
- Appliance control.
Devices such as Google Home and Amazon Alexa are among the most common speech recognition applications.
Medical diagnosis
Another example of ML, helps in the diagnosis of the disease. Many physicians use voice recognition,chatbots to identify patterns in symptoms.
Medical diagnostics examples:
- Oncology and pathology use machine learning to recognise cancerous tissue.
- Assisting in formulating a diagnosis or recommends a treatment option.
- Analyse bodily fluids.
On balance, for rare diseases, facial recognition software is combined with machine learning to scan patient photos and identify phenotypes that correlate with rare genetic diseases.
Thank you for reading Different Types of Machine Learning with Examples. We will conclude this article now.
Different Types of Machine Learning with Examples Conclusion
So far, we have talked about three different types of machine learning, but it is important to note that sometimes the differences between them are not so clear cut or they look almost the same. For example, consider a recommender system. We know this as an unsupervised learning task. It is also easily rephrased as a supervised task. For essence, you just need to label your data.
Summing up, all three types of machine learning aim to teach computers algorithms that allow them to perform tasks more efficiently.
