Glossary term
Glossary term
Foundations
The sum of all the relevant input values multiplied by their corresponding weights. For example, suppose the relevant inputs consist of the following:
The weighted sum is therefore:
weighted sum = (2)(-1.3) + (-1)(0.6) + (3)(0.4) = -2.0
A weighted sum is the input argument to an activation function.
Created for this library
A search team computes the final ranking score as a weighted sum of relevance, freshness, and engagement signals.
A multi-objective recommender team uses a weighted sum of relevance and diversity scores to produce its final ranking.
An ML team uses a weighted sum of several losses during multi-task training to balance objectives.
Definition source: Google for Developers Machine Learning Glossary | Creative Commons Attribution 4.0 License