List view
We need the competitors to be reliable - no matter what kind of publication we write in the end. The data generator should be able to produce a predictable base signal instead of a random walk. DAGMM should be extended.
Overdue by 7 year(s)•Due by June 14, 2018•1/1 issues closedExtend existing approaches using ensembles, entropy filters etc. and validate that the baselines work.
Overdue by 7 year(s)•Due by June 7, 2018•6/6 issues closedEvaluate the competitors on data sets with different outlier types and properties to spot weaknesses.
Overdue by 7 year(s)•Due by May 31, 2018•10/10 issues closed