Metric Dependent

depends on the algorithm used

• 1 / distance measure/metric formula?

Discrete Similarity Metric

anything

• $d(x,y)=\{\begin{array}{ll}\text{1,}& \text{if x=y}\\ \text{0,}& \text{otherwise}\end{array}$

Kernel Functions

vectors

• measures the similarity between 2 vectors in transformed space

Dot Product
Scalar Product

vectors

• 𝛴_{1≤𝑖≤𝑛}(𝑣̅_𝑖𝑢̅_𝑖)

Cosine Similarity

vectors

• 𝑐𝑜𝑠(𝜃) = [𝑢̅·𝑣̅] / [||𝑢̅||*||𝑣̅||]
• used for measuring the similarity of vectors when the magnitude of the vectors does not matter

Jaccard Similarity

sets

• 𝐽(𝐴,𝐵) = |𝐴⋂𝐵| / |𝐴⋃𝐵|