What is Machine Learning?

Machine learning is the subset of artificial intelligence (AI) focused on algorithms that learn patterns from training data and make accurate inferences about new data. This pattern recognition ability enables machine learning models to make decisions or predictions without explicit, hard-coded instructions.

The machine learning definition centers on a simple premise: optimize a model's performance on a dataset of tasks that resemble real-world problems through model training, and the model can make accurate predictions on new data it sees in its ultimate use case. Training is a means to an end. Generalization, the translation of strong performance on training data to useful results in real-world scenarios, is the fundamental goal of machine learning.

Machine learning has come to dominate the field of AI. It provides the backbone of most modern AI systems, from forecasting models to autonomous vehicles to large language models (LLMs) and other generative AI tools.

The term itself is often attributed to Arthur L. Samuel's 1959 article in IBM Journal, "Some Studies in Machine Learning Using the Game of Checkers." In the paper's introduction, Samuel articulates machine learning's ideal outcome: "a computer can be programmed so that it will learn to play a better game of checkers than can be played by the person who wrote the program."

How Machine Learning Works

Machine learning works through mathematical logic. The relevant characteristics, or features, of each data point must be expressed numerically so the data can be fed into an algorithm that learns to map a given input to the desired output.

Data points in machine learning are usually represented in vector form. Each element, or dimension, of a data point's vector embedding corresponds to its numerical value for a specific feature. For data modalities that are inherently numerical, such as financial data or geospatial coordinates, this is straightforward. Many data modalities, such as text, images, social media graph data or app user behaviors, are not inherently numerical. These entail less immediately intuitive feature engineering to be expressed in an ML-ready way.

The process of choosing which aspects of data to use in machine learning algorithms is called feature selection. Feature extraction techniques refine data down to only its most relevant, meaningful dimensions. Both are subsets of feature engineering, the broader discipline of preprocessing raw data for use in machine learning. One notable distinction of deep learning is that it typically operates on raw data and automates much of the feature engineering, or at least the feature extraction, process. This makes deep learning more scalable, albeit less interpretable, than traditional machine learning.

Model Parameters and Optimization

Consider a simple linear regression algorithm for predicting home sale prices based on a weighted combination of three variables: square footage, age of house and number of bedrooms. Each house is represented as a vector embedding with three dimensions: [square footage, bedrooms, age]. A 30-year-old house with 4 bedrooms and 1,900 square feet could be represented as [1900, 4, 30], though for mathematical purposes those numbers might first be scaled, or normalized, to a more uniform range.

The algorithm is a straightforward mathematical function:

Price = (A square footage) + (B number of rooms) - (C * age) + Base Price

Here, A, B and C are the model parameters. Adjusting them adjusts how heavily the model weighs each variable. The goal of machine learning is to find the optimal values for such model parameters: the parameter values that result in the overall function outputting the most accurate results. Most real-world instances of machine learning involve more complex algorithms with a greater number of input variables. The principle remains the same: optimizing the algorithm's adjustable parameters to yield greater accuracy.

Types of Machine Learning

All machine learning methods fall into one of three distinct learning paradigms: supervised learning, unsupervised learning or reinforcement learning. These categories are based on the nature of their training objectives and, often but not always, by the type of training data they entail.

Supervised Learning

Supervised learning trains a model to predict the correct output for a given input. It applies to tasks that require some degree of accuracy relative to some external ground truth, such as classification or regression.

Regression models predict continuous values, such as price, duration, temperature or size. Examples of traditional regression algorithms include linear regression, polynomial regression and state space models.

Classification models predict discrete values, such as the category a data point belongs to, a binary decision or a specific action to be taken. Examples of traditional classification algorithms include support vector machines (SVMs), Naive Bayes and logistic regression.

Many supervised ML algorithms can be used for either task. The output of what's nominally a regression algorithm can subsequently inform a classification prediction.

To be measured and optimized for accuracy, a model's outputs must be compared to a ground truth: the ideal or correct output for any given input. In conventional supervised learning, that ground truth is provided by labeled data. An email spam detection model is trained on a dataset of emails that have each been labeled as SPAM or NOT SPAM. An image segmentation model is trained on images in which every individual pixel has been annotated by its classification. The goal of supervised learning is to adjust the model's parameters until its outputs consistently match the ground truth provided by those labels.

Essential to supervised learning is the use of a loss function that measures the divergence, or loss, between the model's output and the ground truth across a batch of training inputs. The objective of supervised learning is defined mathematically as minimizing the output of a loss function. Once loss has been computed, various optimization algorithms, most of which involve calculating the derivatives of the loss function, are used to identify parameter adjustments that will reduce loss.

Because this process traditionally requires a human in the loop to provide ground truth in the form of data annotations, it's called supervised learning. As such, the use of labeled data was historically considered the definitive characteristic of supervised learning. On the most fundamental level, the hallmark of supervised learning is the existence of some ground truth and the training objective of minimizing the output of a loss function that measures divergence from it.

Unsupervised Learning

Unsupervised machine learning algorithms discern intrinsic patterns in unlabeled data, such as similarities, correlations or potential groupings. They're most useful in scenarios where such patterns aren't necessarily apparent to human observers. Because unsupervised learning doesn't assume the preexistence of a known correct output, they don't require supervisory signals or conventional loss functions.

Most unsupervised learning methods perform one of the following functions:

Clustering algorithms partition unlabeled data points into clusters, or groupings, based on their proximity or similarity to one another. They're typically used for tasks like market segmentation or fraud detection. Prominent clustering algorithms include K-means clustering, Gaussian mixture models (GMMs) and density-based methods such as DBSCAN.

Association algorithms discern correlations, such as between a particular action and certain conditions. E-commerce businesses such as Amazon use unsupervised association models to power recommendation engines.

Dimensionality reduction algorithms reduce the complexity of data points by representing them with a smaller number of features, that is, in fewer dimensions, while preserving their meaningful characteristics. They're often used for preprocessing data, as well as for tasks such as data compression or data visualization. Prominent dimensionality reduction algorithms include autoencoders, principal component analysis (PCA), linear discriminant analysis (LDA) and t-Distributed Stochastic Neighbor Embedding (t-SNE).

Unsupervised learning algorithms can be broadly understood as somewhat optimizing themselves. This animation demonstrates how a k-means clustering algorithm iteratively optimizes the centroid of each cluster on its own. The challenge of training unsupervised models therefore focuses on effective data preprocessing and properly tuning hyperparameters that influence the learning process but are not themselves learnable, such as the learning rate or number of clusters.

Reinforcement Learning

Whereas supervised learning trains models by optimizing them to match ideal exemplars and unsupervised learning algorithms fit themselves to a dataset, reinforcement learning (RL) models are trained holistically through trial and error. They're used prominently in robotics, video games, reasoning models and other use cases in which the space of possible solutions and approaches are particularly large, open-ended or difficult to define. In RL literature, an AI system is often referred to as an agent.

Rather than the independent pairs of input-output data used in supervised learning, reinforcement learning operates on interdependent state-action-reward data tuples. Instead of minimizing error, the objective of reinforcement learning is optimizing parameters to maximize reward.

A mathematical framework for reinforcement learning is built primarily on the following components:

The state space contains all available information relevant to decisions that the model might make. The state typically changes with each action that the model takes.

The action space contains all the decisions that the model is permitted to make at a moment. In a board game, for instance, the action space comprises all legal moves available at a given time. In text generation, the action space comprises the entire vocabulary of tokens available to an LLM.

The reward signal is the feedback, positive or negative, typically expressed as a scalar value, provided to the agent as a result of each action. The value of the reward signal could be determined by explicit rules, by a reward function, or by a separately trained reward model.

A policy is the thought process that drives an RL agent's behavior. Mathematically speaking, a policy (π) is a function that takes a state (s) as input and returns an action (a): π(s)→a.

In policy-based RL methods like proximal policy optimization (PPO), the model learns a policy directly. In value-based methods like Q-learning, the agent learns a value function that computes a score for how good each state is, then chooses actions that lead to higher-value states. Consider a maze: a policy-based agent might learn "at this corner, turn left," while a value-based agent learns a score for each position and simply moves to an adjacent position with a better score. Hybrid approaches, such as actor-critic methods, learn a value function that's then used to optimize a policy.

In deep reinforcement learning, the policy is represented as a neural network.

Deep Learning

Deep learning employs artificial neural networks with many layers, hence deep, rather than the explicitly designed algorithms of traditional machine learning. Though neural networks were introduced early in the history of machine learning, it wasn't until the late 2000s and early 2010s, enabled in part by advancements in GPUs, that they became dominant in most subfields of AI.

Loosely inspired by the human brain, neural networks comprise interconnected layers of neurons, or nodes, each of which performs its own mathematical operation called an activation function. The output of each node's activation function serves as input to each of the nodes of the following layer and so on until the final layer, where the network's final output is computed. Crucially, the activation functions performed at each node are nonlinear, enabling neural networks to model complex patterns and dependencies.

Each connection between two neurons is assigned a unique weight: a multiplier that increases or decreases one neuron's contribution to a neuron in the following layer. These weights, along with unique bias terms added to each neuron's activation function, are the parameters to be optimized through machine learning.

The backpropagation algorithm enables the computation of how each individual node contributes to the overall output of the loss function, allowing even millions or billions of model weights to be individually optimized through gradient descent algorithms. Because of the volume and granularity of updates required to achieve optimal results, deep learning requires very large amounts of data and computational resources compared to traditional ML. That distributed structure affords deep learning models their incredible power and versatility. Essentially, traditional machine learning aims to find a single curve that runs through every data point on a graph. Deep learning pieces together an arbitrary number of smaller, individually adjustable lines to form the desired shape. Neural networks are universal approximators: it has been theoretically proven that for any function, there exists a neural network arrangement that can reproduce it.

Just because something is theoretically possible doesn't mean it's practically achievable through existing training methods. For many years, adequate performance on certain tasks remained out of reach even for deep learning models. Over time, modifications to the standard neural network architecture have unlocked new capabilities for ML models.

Convolutional neural networks (CNNs) add convolutional layers to neural networks. In mathematics, a convolution is an operation where one function modifies, or convolves, the shape of another. In CNNs, convolutional layers are used to extract important features from data by applying weighted filters. CNNs are primarily associated with computer vision models and image data, but have a number of other important use cases.

Recurrent neural networks (RNNs) are designed to work on sequential data. Whereas conventional feedforward neural networks map a single input to a single output, RNNs map a sequence of inputs to an output by operating in a recurrent loop in which the output for a given step in the input sequence serves as input to the computation for the following step. In effect this creates an internal memory, called the hidden state, that allows RNNs to understand context and order.

Transformer models, first introduced in 2017, are largely responsible for the advent of LLMs and other pillars of generative AI, achieving state-of-the-art results across most subdomains of machine learning. Like RNNs, transformers are ostensibly designed for sequential data, but clever workarounds have enabled most data modalities to be processed by transformers. The unique strength of transformer models comes from their innovative attention mechanism, which enables the models to selectively focus on the parts of the input data most relevant at a specific moment in a sequence.

Mamba models are a relatively new neural network architecture, first introduced in 2023, based on a unique variation of state space models (SSMs). Like transformers, Mamba models provide an innovative means of selectively prioritizing the most relevant information at a given moment. Mamba has recently emerged as a rival to the transformer architecture, particularly for LLMs.

Applications of Machine Learning

Most applications of machine learning fall into one or more of the following categories, which are defined primarily by their use cases and the data modalities they operate upon.

Computer Vision

Computer vision is the subdomain of AI concerned with image data, video data and other data modalities that require a model or machine to see, from healthcare diagnostics to facial recognition to self-driving cars. Notable subfields of computer vision include image classification, object detection, image segmentation and optical character recognition.

Natural Language Processing

Natural language processing (NLP) is the subdomain of AI concerned with text and speech data. Machine learning models power NLP tasks such as sentiment analysis, machine translation, text summarization and question answering. Large language models represent the current state of the art in NLP, enabling generative AI applications that produce human-like text.

Predictive Analytics

Predictive analytics uses machine learning to forecast future outcomes based on historical data. Businesses apply predictive models for demand forecasting, customer churn prediction, credit scoring and fraud detection. These applications demonstrate data-driven decision automation in practice.

Recommendation Systems

Recommendation systems use machine learning to predict what products, content or services a user will engage with. Streaming platforms, e-commerce sites and social media networks rely on these systems to personalize user experiences and drive engagement.

Autonomous Systems

Autonomous systems, including self-driving vehicles, drones and robotic process automation, use machine learning to perceive their environment, make decisions and take actions without human intervention. Reinforcement learning plays a significant role in training these systems.

Benefits and Challenges of Machine Learning

Machine learning offers several advantages over rules-based AI and traditional programming approaches.

Benefits

Machine learning systems adapt to new data and improve over time through iterative model refinement. Unlike expert systems that require explicit programming of every rule, machine learning models learn patterns directly from data. This makes them more flexible and scalable as tasks grow in complexity.

Machine learning enables predictive modeling without assumptions about underlying data distributions. Models discover relationships in data that human analysts might miss, supporting data mining for insights and algorithmic pattern discovery.

Automation of feature extraction and pattern recognition reduces the manual effort required to build and maintain AI systems. Deep learning models, in particular, automate much of the feature engineering process, making them well-suited to complex data modalities like images and text.

Machine learning powers AI automation across industries, from healthcare to finance to manufacturing. Organizations use ML to streamline operations, reduce costs and create new products and services.

Challenges

Machine learning models require large amounts of high-quality training data to achieve good performance. Data collection, labeling and preprocessing can be expensive and time-consuming. Biased or unrepresentative data leads to biased models.

Deep learning models are often described as black boxes because their internal decision-making processes are difficult to interpret. This lack of interpretability poses challenges for regulated industries and high-stakes applications where explainability matters.

Training large models demands significant computational resources, including GPUs and specialized infrastructure. This creates barriers to entry for smaller organizations and raises concerns about energy consumption.

Machine learning operations (MLOps) involves managing the lifecycle of ML models, from data preparation to deployment to monitoring. Building robust MLOps pipelines requires specialized skills and tooling.

Models trained on historical data can become outdated as real-world conditions change, a phenomenon known as model drift. Continuous monitoring and retraining are necessary to maintain performance.

Machine Learning vs. Artificial Intelligence

Though machine learning and artificial intelligence are often used interchangeably, they are not quite synonymous. All machine learning is AI, but not all AI is machine learning.

In the popular imagination, AI is usually associated with science fiction, typically through depictions of what's more properly called artificial general intelligence (AGI), like HAL 9000 in 2001: A Space Odyssey or Ava in Ex Machina, or, more recently, with generative AI. But artificial intelligence is a catch-all term for any program that can use information to make decisions or predictions without active human involvement.

The most elementary AI systems are a series of if-then-else statements, with rules and logic programmed explicitly by a data scientist. At the simplest level, even a rudimentary thermostat is a rules-based AI system. When programmed with simple rules like IF room_temperature < 67, THEN turn_on_heater and IF room_temperature > 72, THEN turn_on_air_conditioner, the thermostat is capable of autonomous decision-making without further human intervention. At a more complex level, a large and intricate rules-based decision tree programmed by medical experts could parse symptoms, circumstances and comorbidities to aid diagnosis or prognosis.

Unlike in expert systems, the logic by which a machine learning model operates isn't explicitly programmed. It's learned through experience. Consider a program that filters email spam: rules-based AI requires a data scientist to manually devise accurate, universal criteria for spam. Machine learning requires only the selection of an appropriate algorithm and an adequate dataset of sample emails. In training, the model is shown sample emails and predicts which are spam. The error of its predictions is calculated, and its algorithm is adjusted to reduce error. This process is repeated until the model is accurate. The newly trained ML model has implicitly learned how to identify spam.

As the tasks an AI system is to perform become more complex, rules-based models become increasingly brittle. It's often impossible to explicitly define every pattern and variable a model must consider. Machine learning systems have emerged as the dominant mode of artificial intelligence because implicit learning patterns from the data itself is inherently more flexible, scalable and accessible.

Frequently Asked Questions

What is the simple definition of machine learning?

Machine learning is a subset of AI where algorithms learn patterns from data and make predictions or decisions without explicit programming.

What are the main types of machine learning?

There are 3 main types of machine learning: supervised learning, unsupervised learning and reinforcement learning. Deep learning is a subset of machine learning that uses artificial neural networks with many layers.

How does machine learning differ from traditional programming?

Traditional programming requires explicit rules coded by developers. Machine learning models learn patterns from training data and generalize to new inputs, enabling predictive analytics without rules.

What is the goal of machine learning?

The goal of machine learning is generalization: translating strong performance on training data to useful results in real-world scenarios through accurate predictions and inferences.

What are examples of machine learning in everyday life?

Examples of machine learning include email spam filtering, product recommendations on e-commerce sites, voice assistants, facial recognition and fraud detection systems.

What skills does a data scientist need for machine learning?

A data scientist needs skills in statistics, programming languages such as Python, machine learning algorithms, data preprocessing and model evaluation. Familiarity with frameworks like TensorFlow, PyTorch, Scikit-learn and Keras is valuable.

How do IBM tools support machine learning?

IBM Watson and IBM tech provide platforms for building, deploying and managing machine learning models. The IBM community offers resources for practitioners, and the Think Newsletter delivers industry news on AI, automation, data, quantum, infrastructure and security.

For teams looking to put these ideas into practice, AI agent development can turn machine learning concepts into working systems.