Incremental Isomap

The Incremental Isomap algorithm [28] seeks to provide an efficient solution to the incremental learning problem when using the Isomap algorithm. Incremental 

Isomap has three main steps recomputing the geodesic distances obtaining the lowdimensional representation of the new points and updating the low-dimensional embedding. 

The second step is to find the low-dimensional representation, y, of the new data point X. This can be done using either the Nystrom extension method (Sect. 5.2.1), or using a least squares method such that 

Compute Z = FnewX and perform QR decomposition such that Z = QR. 

Form the matrix Z = Q FnewQ and perform eigen decomposition of the x matrix Z.  

---