How Brunn-Minkowski inequality is used in Machine Learning part3

<ol> <li>Brunn-Minkowski inequality for &theta;-convolution bodies via Ball&rsquo;s bodies(arXiv)</li> </ol> <p>Author : Brunn-Minkowski inequality for &theta;-convolution bodies via Ball&rsquo;s bodies</p> <p>Abstract : We consider the problem of finding the best function &phi;n:[0,1]&rarr;R such that for any pair of convex bodies K,L&isin;Rn the following Brunn-Minkowski type inequality holds</p> <p>|K+&theta;L|1n&ge;&phi;n(&theta;)(|K|1n+|L|1n),</p> <p>where K+&theta;L is the &theta;-convolution body of K and L. We prove a sharp inclusion of the family of Ball&rsquo;s bodies of an &alpha;-concave function in its super-level sets in order to provide the best possible function in the range (34)n&le;&theta;&le;1, characterizing the equality cases</p> <p>2.Brunn-Minkowski inequalities for path spaces on Riemannian surfaces (arXiv)</p> <p>Author :&nbsp;<a href="https://arxiv.org/search/?searchtype=author&amp;query=Assouline%2C+R" rel="noopener ugc nofollow" target="_blank">Rotem Assouline</a></p> <p>Abstract : We define Minkowski summation with respect to a path space on a manifold, extending the well-known notion of geodesic Minkowski sum. For path spaces on two-dimensional Riemannian manifolds consisting of constant-speed curves, we give necessary and sufficient conditions under which this Minkowski summation satisfies a local Brunn-Minkowski inequality.</p> <p><a href="https://medium.com/@monocosmo77/how-brunn-minkowski-inequality-is-used-in-machine-learning-part3-91dcb61e9041"><strong>Visit Now</strong></a></p>