A vector refers to a one-dimensional array of numbers that is used to represent input or output data. Eg. [0.1,0.2,0.15,0.36,0.92].
Vectors are commonly used in neural networks because they can efficiently represent complex data, such as images, audio, and text. Each element in the vector represents a feature or characteristic of the data, such as pixel intensity in an image or frequency in an audio signal.
The values in the vector are typically determined by a feature extraction process, which involves transforming the raw data into a format that the neural network can process. This may involve techniques such as normalization, scaling, or one-hot encoding.
Usually the numbers in the array are between 0 and 1 as this makes the mathematical equations easier to work with and understand. For vector arrays representing an image, each element may represent the intensity of a pixel in the image, with values ranging from 0 to 255, with a vector array representing text data, each element may represent the frequency or presence of a particular word or phrase in the text, with values ranging from 0 to 1 or -1 to 1.