ADOBE REAL-TIME CDP
What is machine learning?
Quick definition: Machine learning is a subset of artificial intelligence (AI), a system that allows computers to perform tasks better and faster than humans, involving models that improve in performance when given more data.
While machine learning (ML) might sound like advanced technology that’s out of reach for many, it’s a surprisingly accessible tool for many organisations. In fact, most people use ML in their daily lives, often without realising it.
Key takeaways:
- ML models improve by “learning” from outcomes that result from certain decisions.
- ML has three key areas — supervised learning, unsupervised learning and reinforcement learning.
- The best time to use ML is when you have more data than humans have the brain power or time to sit through.
- ML enhances productivity by taking over the more tedious tasks of a data analyst’s job.
Consult this guide to learn how machine learning works, the types of ML, the pros and cons and the future of this technology.
Explore the sections of this page
How machine learning works
Types of machine learning
The benefits of using machine learning systems
The drawbacks to machine learning algorithms
How businesses are using machine learning
The history of machine learning
How machine learning works
ML is a type of technology that helps computer systems learn and improve their outputs. Machine learning uses algorithms, which are rules for problem-solving, to generate more helpful creations from machines.
Most ML models work by inputting data into an algorithm. Next, the model will automatically look for errors in its predictions. It uses past examples to compare its outputs and check for any issues. From there, a human user will either accept or reject these outputs. By training the machine learning model, the model becomes more effective and accurate over time. As ML models gather more data and experience, they require less human intervention.
ML is similar to but different from AI. Machine learning is the process of a computer becoming more intelligent by iterating over time. But with AI, the computer uses its knowledge to perform tasks without human intervention. The biggest difference between these two technologies is that AI can mimic human intelligence, while ML simply performs tasks based on pattern recognition.
Types of machine learning
Supervised learning
Supervised learning is a type of data science that uses labelled data that’s been tagged with specific information about outcomes associated with that data. A model is then trained to learn what features or variables are predicting the outcomes assigned to the labelled input data. The model can then use the information from the output data to assess its own performance and predict outcomes.
Supervised learning involves two main use cases — classification and linear regression. Classification predicts a class label. For example, you might predict whether a customer will cut ties with a brand based on attributes such as purchase behaviour.
Linear regression predicts a numerical label, like the expected revenue you think you’ll receive from a customer based on specific attributes. The outcome is a numerical variable, as opposed to a condition.
Supervised learning requires a solid machine learning model to generate quality outputs. Learn how ML models affect the different types of outputs a computer can generate.
Unsupervised learning
Unsupervised learning starts with a set of raw, unlabeled data. The main purpose of unsupervised learning is to find connections between the dataset and any additional data points you give the model.
This method can help you to find relationship-based groups within data or clusters, which can be used to create customer segments.
Reinforcement learning
Reinforcement learning starts by inputting a set of raw, unlabeled data into a model. The model then takes action. Based on that action, the model receives feedback on whether it acted correctly or incorrectly and the outcomes of that action. The model then creates another action and keeps learning until model optimisation is achieved.
A great real-world example of reinforcement learning is a recommendation algorithm on a film streaming service like Netflix. The service shows you a film you may or not like and learns from your “like” or “dislike” rating to determine whether it should keep recommending the same types of films.
The history of machine learning
While ML might sound like new technology, it’s been around for decades. The machine learning we know today has its roots as far back as the 1940s.
1940s
In 1943, Warren McCulloch and Walter Pitts created the first neural network. This allowed computers to communicate with each other without human interaction.
1950s
Alan Turing created the Turing Test to determine whether machines can behave like humans. Researchers still use the Turing Test today to see if humans can tell the difference between human-generated and machine-generated outputs.
1960s
Thomas Cover and Peter Hart published the K-Nearest Neighbours (KNN) algorithm, one of the first ML algorithms that could identify patterns from a large amount of data.
1970s
Paul Werbos wrote a dissertation in 1974 titled The Roots of Backpropagation, which paved the way for backpropagation — a technology that allows neural networks to recognise patterns more accurately.
1980s
Explanation Based Learning (EBL) made it possible for computers to analyse and train themselves on data, as well as disregard unimportant data. Artificial neural network NetTalk also learnt how to correctly pronounce English text.
1990s
In 1997, IBM stunned the world when its supercomputer Deep Blue defeated an expert human chess player. This showed the world that machine learning could match and even exceed human performance.
2000s
Torch (now known as PyTorch), a free software library, became the world’s first large-scale ML platform — making machine learning much more accessible. During the 2000s, computers also learnt how to “see” text and images with deep learning.
2010s
Google develops Google Brain, a deep neural network that automatically categorises objects. Facebook, Amazon and Microsoft also develop ML models.
2020s
In November 2022, OpenAI’s ChatGPT took the world by storm. This technology made ML and AI accessible to everyday people, who use the technology for everything from generating job cover letters to writing emails.
The benefits of using machine learning systems
Machine learning is a useful technology that helps businesses:
- Enhance productivity. More than 80% of employees believe AI improves their performance at work. By automating tasks normally reserved for humans, ML can increase organisational productivity.
- Help customers. Companies use ML to optimise their products to make their customers’ lives easier. In fact, 62% of consumers are willing to share their data if it means they will have a better experience.
- Reduce human error. ML automates manual tasks that are normally prone to human errors and typos. This means your organisation can leverage cleaner data for better business insights.
- Enhance availability. Consider that 51% of consumers expect businesses to be available 24/7. Implementing ML solutions for your business makes you available to customers day and night.
- Remove risk. Falling out of compliance can cost your organisation millions in fines and lost business. Fortunately, ML follows predetermined rules to help you stay compliant at scale.
- Reduce repetitive tasks. ML automates tedious tasks, like data entry, so your employees can focus on more high-value tasks. In fact, 68% of employees want more AI-based technologies to help them to execute tasks.
- Uncover insights. ML can uncover insights that humans often don’t have the brain power to even consider.
The drawbacks to machine learning algorithms
There are a few drawbacks to using machine learning systems, including:
- Amount of data required. ML models often require a large amount of data to be effective.
- Size of dataset. The size and quality of a dataset are two of the biggest factors in determining how good a model is and the more data you have, the more time it takes to label that data accurately for use in supervised learning methods.
- Combining data types. You need to think about the different types of data that you need to add to your dataset to make the model a robust one because you’re teaching the machine to make decisions the way that humans would.
- Potential bias. Another drawback of ML concerns ethics, especially when it comes to deep learning. Many of these models don’t share how they make decisions, so you’re not exactly sure what factors they are using. The model is only as good as the data that you feed into it, but you still might not know what kinds of relationships the model will notice and whether they’re morally fair.
- Potential inaccuracies. It’s important to remember that while it can be more effective than human outputs, ML isn’t infallible. If there are inaccuracies in your data or logic, the machine learning model will reflect those inaccuracies.
- Cost. If you want to create a custom ML solution for your organisation, there are costs associated with hiring data scientists to build and maintain these models. On average, organisations spend $60,000 to $95,000 over the first five years of using a model. However, opting for solutions like Adobe Sensei allow your organisation to offset many of these costs.
Machine learning isn’t perfect, but organisations can overcome many of the drawbacks by choosing the right ML model for the right scenarios.
How businesses are using machine learning
There are many instances where implementing ML algorithms can help streamline and optimise an organisation’s resources. One common case is when the big data is too much for a human to sift through, but it holds important information that could inform company decisions.
This benefit isn’t reserved for just technology companies, either — 68% of companies worldwide use ML and that number will likely increase. Because machine learning is so beneficial, its use is expanding across different industries.
Healthcare
Healthcare companies use ML to process a lot of patient data — all while complying with data protection laws. For example, some organisations use machine learning to improve the accuracy of diagnostic imaging to detect diseases in their earlier stages. It’s also helpful for detecting fraud, spotting errors and personalising treatment.
Manufacturing
More manufacturers are embracing ML to do smarter preventive maintenance. Instead of wasting time and money maintaining machines that don’t need maintenance just yet, manufacturers use machine learning to analyse data trends and optimise which machines are serviced.
But this isn’t the only way manufacturers use machine learning. They also take advantage of this technology to manage semi-autonomous and autonomous vehicles within their facilities.
Entertainment
Have you got a Netflix account? If so, you’ve seen ML in practice. The streaming service uses machine learning to personalise thumbnails, recommend films and shows and optimise streaming quality.
Marketing
Let’s say your marketing team is reviewing new data from different customer segments. It’s difficult to sort through the different attributes of customers to identify high-performing segments or what the preferences of Customer Segment An are as opposed to Customer Segment B.
With the amount of data available, it’s not likely your marketing team will have the brain power to process it all, let alone gather any useful insights. This is something ML can do quickly and likely better than humans.
Using machine learning automation could result in predictive insights your team might not have found on its own.
Finance
Robo-advisers are becoming much more popular in the world of finance. For example, services like Wealthfront use ML and AI to automate portfolio management for their users. This reduces the amount of time that finance companies spend managing clients’ investments while still giving clients the benefit of managed investments.
Power incredible customer experiences with machine learning.
Need Improved segmentation and personalisation with machine learning (ML)? Adobe Real-Time CDP can help.
Recommended articles
https://main--bacom--adobecom.hlx.page/fragments/products/cards/rtcdp
