what are weights in a machine learning model
In: 7
Lets imagine that you want to find a model that tries to predict the weather tomorrow based on the reading of different instruments in you home, e.g. you have a thermometer, a barometer, a counter measuring current percipitation im mm, a sensor measuring sunlight intensity, a windhose etc.
Each of these inputs may or may not affect your prediction and different factors should have different importance for your prediction. Therefore “weights” are introduced to give each factor an individual emphasis. In a very simple model, the readings of all your instruments are just multiplied by their respective weights and the products summed up to give a scoring function. In a more complex model more complex calculations are done, but the principle idea stayes the same.
But how do you know how to weight different factors? This is what you find out during training. Here you adjust all the different weights until it gives you the most reliable prediction.
Latest Answers