Cosine similarity between two matrices

33 ビュー (過去 30 日間)
Paola 2022 年 5 月 24 日
コメント済み: Paola 2022 年 5 月 30 日
Dear all,
I have some vectors 32x1, representing force fields. I use the quiver function to plot and visualize the fields. (see pic attached) Obtaining a quiver showing 4x4 vectors, for a total of 16 vectors.
X = [1 1 1 1 2 2 2 2 3 3 3 3 4 4 4 4];
Y = [1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4];
U = W(1:2:31);
V = W(2:2:32);
figure, quiver(X,Y,U',V');
Even if visually they look very similar, I need to calculate a cosine similarity value, between the different vectors.
Checking online I found that this formula:
cosSim = dot(a,b)/(norm(a)*norm(b));
and also the function
D = pdist(X,'cosine').
I wonder if these two functions can be applied to my case and which one is better to obtain just a similarity value for each couple of vector compared.
Thank you!

サインインしてコメントする。

採用された回答

Matt J 2022 年 5 月 24 日
I tend to prefer pdist2,
UV=[U(:),V(:)];
D=1-pdist2(UV,UV,'cosine')
5 件のコメント表示非表示 4 件の古いコメント
Paola 2022 年 5 月 30 日
Thank you!

サインインしてコメントする。

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by