What is Machine Learning?
As of 2020, we’re all pretty familiar with the idea of robots – physical and virtual – designed to make our lives easier. These robots are typically pre-programmed to follow set scripts in order to carry out specific tasks.
But do you know what happens when computer systems start to not only execute pre-set programs but actually absorb information and starting becoming smarter by themselves? That’s called machine learning.
But does “machine learning” really mean that these machines are… learning?
Well yes, sort of. Although we use the term “learning” for this type of computing, it’s not about machines becoming smarter in a cognitive sense. This learning is actually based on the processing of data and identifying patterns that may predict future scenarios – but these predictions will only ever be as good as the data that is being analyzed. As such, it doesn’t necessarily mean the computer will help people make better decisions.
What about Artificial Intelligence? Isn’t it the same thing?
Not really, while similar there are some key differences between artificial intelligence and machine learning. When machine learning first was designed, it was considered to be a stepping stone on the journey towards Artificial Intelligence (AI).
However, as AI developed, it became gradually clear that they were two distinct strands of technology. Traditional machine learning was dependent on having enough (and accurate) data available to draw statistical conclusions from it, whereas AI was about emulating the logical decision-making of a human being. This is why the field of machine learning changed to focus on problem-solving using different statistics.
In other words, Machine learning is actually an application of artificial intelligence (AI) that provides systems the ability to automatically learn and improve from their experience without being explicitly programmed. Machine learning focuses on the development of computer programs that can access data and use it to learn for themselves.
A few examples of machine learning
Machine learning algorithms are often categorized as supervised or unsupervised.
Supervised machine learning algorithms can apply what has been learned in the past to new data using labeled examples to predict future events. They start by analyzing a training dataset, after which the learning algorithm produces an inferred function to make predictions about the output values. The system is able to provide targets for any new input after sufficient training. The learning algorithm can also compare its output with the correct, intended output and find errors in order to modify the model accordingly.
On the other hand, unsupervised machine learning algorithms are a bit different because these are used when the information used to train is neither classified nor labeled. Unsupervised learning studies how systems can infer a function to describe a hidden structure from unlabeled data. The system doesn’t figure out the right output, but it explores the data and can draw inferences from datasets to describe hidden structures from unlabeled data.
There’s also a third type of ML called “semi-supervised machine learning”, whose algorithms fall somewhere in between supervised and unsupervised learning since they use both labeled and unlabeled data for training. These typically use a rather small amount of labeled data and a large amount of unlabeled data. The systems that use this method are able to considerably improve learning accuracy. Often semi-supervised learning is chosen when the acquired labeled data requires skilled and relevant resources in order to train it or learn from it. Otherwise, acquiring unlabeled data generally doesn’t require additional resources.
And finally, the fourth and final type of machine learning is called “reinforcement machine learning”. This uses algorithms that interact with the environment and produce actions and discover errors or even rewards. In other words, a reinforcement algorithm learns by trial and error to achieve a clear objective. It tries out lots of different things and is rewarded or penalized depending on whether its behaviors help or hinder it from reaching its objective. This method allows machines and software agents to automatically determine the ideal behavior within a specific context in order to maximize its performance. Simple feedback is required for the agent to learn which action is best. It’s quite a complex type of machine learning, but if developed and used properly, it can be extremely useful for large businesses.
How can machine learning be beneficial for us?
Well for beginners, it can take over risky jobs, such as bomb defusals/disposals done by robots. Right now, most of these robots require a human to control them. But as machine learning technology improves in the future, these tasks would be done completely by robots with AI. This technology alone has already saved thousands of lives.
It can even help the elderly, as for many seniors, everyday tasks can be a struggle. Many have to hire outside help or rely on family members. But a robot that can do most of these chores, and automatically improve itself day by day can be amazing.
It can also be extremely helpful for healthcare. Hospitals may soon put your wellbeing in the hands of an AI, and that’s good news. Hospitals that utilize machine learning to aid in treating patients see fewer accidents and fewer cases of hospital-related illnesses, like sepsis. AI is also tackling some of medicine’s most intractable problems, such as allowing researchers to better understand genetic diseases through the use of predictive models.
Closing thoughts
As you can see, there are many, many advantages of ML, and if used correctly, it could even be life-changing for millions of people. Currently, the most promising approach to artificial intelligence is the use of applied machine learning. Rather than trying to encode machines with everything they need to know upfront (which is impossible), we want to enable them to learn, and then to learn how to learn.