As aptly quoted by Andrew Ng , “AI is the new electricity”. This just goes on to emphasize the versatility and importance of Artificial Intelligence (AI) and Machine Learning (ML) in today’s world. The boom of ML and its applications in all domains, be it finance, healthcare, logistics or education, has shown how it will have an ever increasing role in the future. In this article, we will learn about Machine Learning, understanding what it is, and how to get started with it.
What is Machine Learning?
Machine Learning is a field of Computer Science where a system learns to find patterns in examples in order to make decisions, without being explicitly programmed to do so. These algorithms improve automatically, using the data provided and with experience.
In other words, ML Algorithms are the algorithms that tell us something interesting about the data, find patterns in them, without us having to write any custom code specific to the data. Instead of coding the exact algorithm, we provide data to these algorithms and they automatically learn to find the patterns and build their logic based on the data. ML is what powers Facial Recognition, Google Search prediction, Alexa and Siri, autocorrect, and has utility in almost all domains today.
Machine Learning is a subset of AI (Artificial Intelligence). Artificial Intelligence is a technology which appears to mimic human behavior. In other words, AI depicts behavior that is considered as “smart” or “intelligent”. It includes Machine Learning as well as more traditional, rule-based algorithms that have explicitly written instructions. A great example of AI application is robotics.
Why learn Machine Learning?
Impact of ML in projects and products is very huge nowadays. It has many applications in every field like marketing, operations, e-commerce etc . It allows us to make predictions from data by learning the relationship between features of the data, and finding patterns. Machine learning can be used from predicting housing prices, predicting stocks values, to whether an image contains a cat or a dog, or a human, or person. The applications extend far along to the more complicated areas like biology for understanding gene regulatory networks, or the pinnacle of ML, self-driving cars, to the fraud detection and recommendation systems that make our lives easier.
Due to all these applications, machine learning is also one of the most highly sought after skills of this century. Jobs in Data Science and Machine Learning have salaries exceeding that of traditional IT streams. In the last decades, the volume and varieties of data has grown rapidly. Simultaneously, with Computational processing getting cheaper, more powerful and affordable, analyzing a huge amount of data is leading to very valuable insights . This plays a significant role in decision making of organizations. And all of this is being automated, thanks to Machine Learning !
Types of Machine Learning
Machine Learning Algorithms can be broadly classified into 3 types ( based on what kind of data they learn from), namely : supervised, unsupervised & reinforcement learning.
Supervised Learning: Supervised learning algorithm takes labeled data and creates a model that can make predictions. In supervised learning, the machine learns from the data with the help of labels. In other words, the machine develops decision making capacity with the guidance of labelled data. The learning is what we call training the model, and for that we provide it labelled data. Using these, the model can now predict values for new data (data unknown to it). Examples of Supervised learning models are: Classification, regression, prediction, gradient boosting etc.
Unsupervised Learning: Unsupervised learning is when we are dealing with data that has not been labeled or categorized. The data on which the model trains here is unlabelled, and just has its features without any label. Here, the system has to learn from just the patterns in the data, and from the similarity and differences in different clusters of data.
Examples of Unsupervised Learning models are clustering algorithms such as k-means clustering, nearest neighbor mapping, and self organizing maps. These algorithms also have a huge application since most of the data we deal with today are not cleaned and labelled, and we use them for applications like segmenting text topics, or for recommending items and or identifying the outliers in unlabelled data.
Reinforcement Learning: Reinforcement learning is the type of machine learning which uses a checks and reward system and trial-and-error in order to maximize the long-term reward. In this, the algorithm ventures in the unknown and discovers through trial & error what actions will lead to long-term profits.
These algorithms are used extensively in robotics, gaming and navigation. Since these types of applications contains the primary components of Reinforcement Learning:
1. Agent: The Decision maker.
2. Environment: Everything our agent interacts with.
3. Actions: steps the agent takes, or thing agent can do.
How to get started:
The prerequisites for getting started with Machine Learning are a few, and easy to grab hold of, and the impacts and scope huge, as explained in the previous section. Let us now see how to actually get started with Machine Learning and what we need to learn.
Machine Learning was essentially born out of statistics, and so it makes sense that understanding the essence of ML will require mathematics. Concepts such as Statistics, Linear Algebra and Probability Theory are very important foundations of Machine Learning and a basic understanding and intuition of these topics would suffice to really understand how the models actually work.
But even if you are not from a Mathematics background, there’s no need to panic, since there’s a lot more to it than that, and for most of the basic applications, a basic understanding of intuition can be enough to apply ML to your work.
To learn machine learning, you will have to learn to code.Since machine learning is tied very tightly to data, almost all types of coding languages used for this domain are the ones often used with data modelling or statistics. Theoretically, any programming language can be used for practicing Machine Learning, however, it is normally preferred to use programming languages like Python or MATLAB. This is because data is easier to manipulate with these languages as compared to others. Python is by far the most popular coding language for ML, in both industry and academia. For a better prospect in machine learning, you can learn python online. MATLAB also has a lot of built in support for ML and is used frequently for applications. Both of these are easy and quick to learn.
Implementing ML, the language & frameworks
Thanks to the modern frameworks and libraries today ( For example Tensorflow, keras, theano, Pytorch), implementing a Machine Learning model has become very convenient, and can be done using just a few lines of code. You can use Python or even C++ together with these libraries. For example, the most famous of these, TensorFlow, is free and open-source and is used extensively for machine learning applications such as neural networks, and it abstracts the hundreds of computation units and mathematics behind them to just a few lines of code .
And not just codes for simple models like linear regression or classification, these libraries sometimes provide us with everything from the most important datasets to some really complex models which we can implement using their functions. Also, you can use pre-trained ML models and change some parameters to get the result of your choice on your own data.
In this article, We learned the concept of Machine Learning and Artificial Intelligence and the impact of this technology. We got to know how ML is shaping the world today and why it is one of the most coveted skills. We further discovered the types of Machine Learning and what each of them entails. And finally got to know how to get started with Machine Learning and what languages and tools to use. Stay tuned for more informative articles
Disclosure of Material Connection: Some of the links in the post above are “affiliate links.” This means if you click on the link and purchase the item, I will receive an affiliate commission. Regardless, I only recommend products or services I use personally and believe will add value to my readers.