Artificial Intelligence (AI) has become a buzzword in recent years, mentioned by many but rarely understood. What exactly is AI? And how does it work? In this blog series, we’ll explore the fascinating world of AI, shedding light on its fundamental principles and mechanisms.
Our intellect holds immense significance in our lives. Over millennia, we have dedicated ourselves to unravelling the mysteries of human cognition and behavior. This quest involved understanding how our brains are capable of perceiving, comprehending, predicting, and interacting with an endlessly complex world.
In our previous blog, we explained that Artificial Intelligence (AI) is a broad field that involves creating systems or software capable of performing tasks that typically require human intelligence. We know explore one of its key elements, Machine Learning
The Magic Behind AI: Machine Learning
At the heart of AI is a concept known as machine learning. Machine learning is a subset of AI that uses algorithms to automatically learn and improve from experience. It’s the ability of a machine to adapt to new circumstances and detect and extrapolate patterns. Machine learning is typically categorized into four main types: supervised learning, unsupervised learning, semi-supervised learning, and reinforcement learning.
Supervised Learning
Supervised Learning involves training a model to learn the relationship between input data and corresponding output labels. The algorithm learns by example, gradually grasping patterns and relationships within the data.
Supervised Learning starts by gathering or creating a dataset with input-output pairs, like location, market information (x) and housing prices (y). Next, we introduce this data to our machine learning model, which will try to uncover hidden patterns in the relationships between input and output variables, a process that is called training.
The model’s task is to uncover hidden patterns, essentially playing detective in the relationship between the input and output variables. During this phase, the model fine-tunes itself, narrowing the gap between predictions and actual labels, ultimately learning from its errors, and improving accuracy. In other words, refining its secret sauce.
Once trained, the model faces a fresh challenge: the test dataset. This dataset remains isolated from the rest, unseen until now. It represents uncharted territory, mirroring real-world scenarios the model has never encountered before. Here, we evaluate how well our model generalizes its knowledge, testing its ability to make accurate predictions in entirely new scenarios.
Unsupervised learning
Unsupervised Learning operates on unlabeled data, seeking to discover relationships and hidden structures autonomously. It doesn’t rely on predefined labels, allowing it to reveal previously unnoticed patterns and insights. This approach is particularly useful in scenarios where we aim to extract valuable information from raw data without a clear input-output overview. Through techniques such as clustering and dimensionality reduction, unsupervised learning opens the door to discovering intrinsic patterns, making it a critical asset in the realm of machine learning and data analytics.
Semi-supervised
Semi-Supervised Learning includes techniques that bridges the gap between fully labelled data and unsupervised learning. It trains a model on a dataset with only a portion of the labelled examples alongside another dataset of unlabeled data. Meaning, labelled data is used to provide the model some supervision and guidance while leveraging the larger pool of unlabelled data to improve the model’s performance. This approach is particularly useful when labelled data is scarce or expensive to gather.
Reinforcement learning
Reinforcement Learning is like teaching a computer program to learn from its mistakes and make better decisions over time. It teaches an agent how to choose its actions from an environment, searching for the maximized reward. Reinforcement learning is pretty much training a digital pet to perform tricks, play games and solve problems. Let’s explain that.
Imagine you have a puppy called Luca (agent), and you want to train it to do tricks such as giving a high-five (action). When Luca does a trick correctly in the living room (environment) she gets a treat (reward). Over time, Luca figures out which actions get treats, her desired outcome.
Conclusion
While supervised learning relies on labelled data to generate predictions, unsupervised learning finds logical patterns in unlabelled data. Semi-supervised learning combines labelled and unlabelled data to improve performance typically when labelled data is scarce or expensive to gather, and reinforcement learning focuses on decision-making through interactions of agents within an environment. Depending on the availability of data, desired outcome, or complexity of the learning task you will choose one of them for you AI system or programme. Which one should you choose for your business case?
