Inputs and Features
Distinguish raw input data from the numerical features a model receives.
Core explanation
A model operates on representations of data that can be processed mathematically. In the spam-filter teaching example, the inputs were represented as:
- x1: promotional language present
- x2: suspicious link present
- x3: sender treated as untrusted
For the worked examples, each feature was represented as 1 (present) or 0 (absent).
Important qualification
Binary inputs were chosen to make the arithmetic easy to follow. Real model inputs are not necessarily binary. They may be continuous values, categorical encodings, token IDs, embeddings, images, audio representations or other numerical structures, depending on the model.
Feature names are labels chosen by the designer or data pipeline. The name “suspicious link” does not mean the model inherently understands suspiciousness; it receives a numerical representation.
Exercise context
Use the inputs (1, 1, 1) with weights (0.4, 0.9, 0.7) and bias -0.5 in the neuron example. Compute each feature's contribution before adding the bias.
Knowledge check
1. Must every model input be binary? No. Binary values were a simplification; real inputs can have many numerical forms. 2. Does a feature label prove that a model understands the human concept behind it? No. The label describes the feature representation; it does not establish semantic understanding.
Recovered from the available teaching record; not a verbatim transcript.
Introduced; learner mastery not verified.
Help improve this learning path.
Suggest a concept we should teach, or tell us what is unclear. No account is required. Email is optional and only for a reply. Do not include confidential or sensitive information.