Feature
A feature is a specific piece of information from your data that helps a computer learn, like looking at the square footage of a house to guess its price. Choosing the most helpful pieces of information is a skill called feature engineering.
A feature is an individual measurable attribute or variable used as an input for a machine learning model. Feature engineering is the process of selecting, modifying, or creating these variables from raw data to improve the model's predictive performance.
A feature is a distinct, quantifiable property of an observed phenomenon used as an input vector component for a statistical or machine learning model. Feature engineering is the systematic process of domain-driven transformation and selection of raw data into optimized feature sets to maximize the model's objective function.
evolution
- 1957 · historyPerceptron Input Weights
Frank Rosenblatt introduced the Perceptron, where individual input signals were assigned weights, establishing the foundational concept of features.
- 1990 · historyFeature Selection Formalization
George John and Ron Kohavi formalized the wrapper and filter methods for feature selection, distinguishing between relevant and irrelevant input signals.
- 2006 · historyDeep Learning and Representation Learning
Geoffrey Hinton popularized deep belief networks, shifting the paradigm from manual feature engineering to automated feature extraction via hierarchical layers.
- 2017 · historyAttention Mechanisms
The Transformer architecture introduced self-attention, allowing models to dynamically weigh the importance of different input features based on context.