There are several known contributions of Topology to ML.
But: what has been the contributions of ML to the research on General Topology, and particularly to Finite Topological Spaces? Do you have any information of this regard?
Otherwise, how to solve problems in General Topology (proving theorems, or supporting the prove, obtaining open or close systems of subsets, determining open basis, etc) by application of ML techniques or its theoretical principal (in that last case it looks to be some kind of recursive application, given the heavy dependence of ML methods on metrical space properties considerations).
It's supposed most of those methods or approaches are related to formalization of real world situations, by means of reducing the complexity of analyzed space.