Why are noise and outliers treated as the same concept in DBSCAN (density-based spatial clustering of applications with noise)?
DBSCAN treats everything that cannot be assigned to a cluster under the parameterization given separately. Clusters are signal for clustering algorithms. Everything that is not signal is by definition noise, so everything that DBSCAN cannot cluster is labeled "noise". This applies equally to outliers (points that are so far away from all clusters they cannot be clustered) and to "inlier" (points that lie between clusters, but far away from every one, and therefore cannot be assigned confidently to any one cluster).