Open problems for resistance-Laplacian anomaly detection

Develop improved graph-construction procedures, alternative resistance-based anomaly score functions, normalization schemes, and theoretical guarantees linking the dominant resistance-Laplacian eigenvector to graph-theoretic isolation.

Background

The anomaly-detection experiments are explicitly described as preliminary and proof-of-concept. The proposed score is heuristic, and its performance varies across datasets: it identifies some injected outliers but fails to distinguish another from naturally peripheral points.

The authors explicitly identify several unresolved directions: improving the graph used to represent the data, designing better scoring functions, determining suitable normalization schemes, and proving guarantees connecting the dominant resistance eigenvector with graph-theoretic notions of isolation.

References

Open problems include improved graph construction, alternative score functions, normalization schemes, and theoretical guarantees linking the dominant resistance eigenvector to graph-theoretic isolation.

— Spectral properties of the resistance Laplacian with applications to data clustering and anomaly detection  (2609.04768 - Deshpande et al., 4 Sep 2026) in Section 3.3, subsection “An exploratory anomaly detection experiment”

Note that relying on the MST is a design choice. Since higher-dimensional spaces lack absolute ordering, the selection of the topological backbone is an open question. Other edge-selection strategies could also be employed to define $d_k$.

— RIFT: Relative Isolation From Trees For Anomaly Detection  (2610.12244 - Szalai et al., 8 Oct 2026) in Section 3.1, “Concept of the algorithm”