B-Spline
A piecewise polynomial curve defined by control points, possessing local support such that each segment influences only a limited region. Adjacent segments maintain smooth continuity. B-splines are widely used in computer graphics and served as the function parameterization in the original KAN architecture.
Timeline (last 90 days)
B-splines solved the problem of Bézier curves lacking local modification capability through the local support of basis functions, making piecewise editing of complex curves possible.
The de Boor algorithm provides a guarantee for numerically stable evaluation of B-splines.
The local support of B-spline basis functions means that each control point affects only one interval of the curve, and its range of influence is determined by the knot vector.
By adjusting knot multiplicity, the order of smoothness of a B-spline curve at the joins can be controlled.
The Kolmogorov-Arnold network originates from the Kolmogorov-Arnold representation theorem, placing learnable univariate functions (usually parameterized with B-splines) on the edges; in theory it can approximate complex functions with fewer parameters and is especially good at fitting smooth continuous mappings, but the engineering complexity of spline computation and parameter initialization is higher.
All Facts (5)
B-splines solved the problem of Bézier curves lacking local modification capability through the local support of basis functions, making piecewise editing of complex curves possible.
50%UnverifiedThe de Boor algorithm provides a guarantee for numerically stable evaluation of B-splines.
50%UnverifiedThe local support of B-spline basis functions means that each control point affects only one interval of the curve, and its range of influence is determined by the knot vector.
50%UnverifiedBy adjusting knot multiplicity, the order of smoothness of a B-spline curve at the joins can be controlled.
50%UnverifiedThe Kolmogorov-Arnold network originates from the Kolmogorov-Arnold representation theorem, placing learnable univariate functions (usually parameterized with B-splines) on the edges; in theory it can approximate complex functions with fewer parameters and is especially good at fitting smooth continuous mappings, but the engineering complexity of spline computation and parameter initialization is higher.
50%