Title: Machine Learning Author: Kriko Published: May 30, 2023 Last modified: Jul 7, 2026 --- 1. [Home](https://kriko.io/) / 2. [Glossary](https://kriko.io/glossary) / 3. M Letter # Machine Learning **Machine learning**, commonly abbreviated as ML, is a field of artificial intelligence that enables computer systems to learn patterns from data, generate predictions and improve their performance over time. Instead of defining detailed rules for every possible situation, machine learning uses algorithms, statistical models and training data. The model learns relationships from the examples it receives and attempts to generate results for information it has not previously encountered. Its effectiveness depends on how accurately the training process reflects the problem being addressed. Machine learning and data mining are closely related, but their objectives are not identical. Data mining focuses on discovering patterns, relationships and meaningful information within large datasets. Machine learning uses these patterns to generate predictions, classifications, recommendations or automated decisions. In practice, the two fields may be used together, and the distinction between them can vary according to the project. During a machine learning process, large volumes of data are processed by algorithms to create a model for a defined objective. Model performance depends on factors such as data quality, algorithm selection, feature preparation and the methods used for evaluation. Once training is complete, the model can generate predictions or classifications for new inputs. However, its results should be monitored continuously, and retraining may be required when the underlying data changes. Medical imaging is one example of machine learning in practice. A model may be trained using a large number of images and diagnoses previously labelled by medical professionals. It can learn visual patterns associated with specific conditions and help identify similar findings in new images. However, such systems should support professional judgement rather than replace qualified medical assessment. Machine learning does not completely eliminate the need to write code. Preparing models, processing data, selecting algorithms and evaluating results generally require software development and data science expertise. Although ready-made platforms can simplify certain stages, professional knowledge remains important for managing data quality, model selection and performance. Machine learning is used in areas such as image recognition, search systems, sentiment analysis, fraud detection, demand forecasting and personalised product recommendations. **Supervised learning** uses labelled datasets in which input data is provided together with the correct expected outcome. The model learns the relationship between inputs and outputs and applies this knowledge to new information. Classifying email messages as spam or legitimate is a common example. Regression and classification problems are among the main applications of supervised learning. **Unsupervised learning** focuses on discovering structures and relationships within datasets that do not contain predefined labels. The model may group information with similar characteristics or identify unusual patterns in the data. Customer segmentation and anomaly detection are common applications of this approach. Professional interpretation is often required to determine the meaning and business relevance of the resulting groups or patterns. **Semi-supervised learning** combines a small amount of labelled data with a larger volume of unlabelled information. It can be useful when labelling data is expensive, time-consuming or requires specialist knowledge. The model learns fundamental relationships from the labelled examples while extracting additional structures from the unlabelled data. This method may be applied in image, audio and text processing projects. **Reinforcement learning** is based on a system learning from the results of the actions it performs within an environment. The model receives rewards for successful or useful actions and lower rewards or penalties for unsuccessful ones. Its objective is to develop an effective behaviour strategy by considering both immediate and long-term outcomes. Reinforcement learning is used in robotics, gaming systems, resource management and autonomous decision-making. In summary, machine learning is a powerful artificial intelligence approach that uses patterns in data to support prediction, classification and decision-making. Successful results depend not only on the algorithm but also on data quality, accurate problem definition and continuous model monitoring. The same learning method is not appropriate for every project, and the selected approach should reflect the relevant business objective. When implemented effectively, machine learning can improve operational efficiency and support more informed decisions. ## Discover it in the dictionary ### Pay Per Click (PPC) Pay Per Click, commonly abbreviated as PPC, is a digital advertising model in which advertisers pay when users click on their ads. In this… ### Feedback Feedback is an evaluation, comment or report that a user provides to a system, platform, brand or search engine about an action they have… ### Inbound Marketing Inbound marketing is a marketing approach that aims to attract potential customers to a brand through relevant content, experiences and digital touchpoints instead of… ### Dirty Data Dirty data refers to information that is inaccurate, incomplete, inconsistent, invalid, duplicated or outdated. Data is an important resource for strategic decision-making, but it… ### Lookalike Audience A Lookalike Audience is an advertising targeting method used to reach new people who share similar characteristics with your existing customers, website visitors, app… ### Scrum Scrum is a framework used in agile project management and product development to break complex work into smaller and more manageable parts. Although it… ### Gamification Gamification refers to the use of game mechanics in non-game contexts. This approach integrates elements such as points, badges, levels, missions, rewards, competition, progress… ### Data Literacy Data literacy is defined on Wikipedia as the ability to read, work with, analyse and discuss data. More broadly, it refers to the ability… ### Target Audience Target audience refers to the specific group of people a brand wants to reach through its product, service, campaign or communication activities. This group… ## Track the digital heartbeat with Kriko Subscribe to receive curated insights, news, and ideas shaping the digital landscape. By checking this box, you acknowledge and accept our [privacy policy](https://kriko.io/privacy-policy).