Sharp non-asymptotic minimax rates are derived for submatrix detection in Gaussian matrices, with matching upper and lower bounds on the critical signal strength μ* for all parameter configurations.
Then if µ2 < cµ s2 log (s2d1 log(cd2 s2 ) cµ2es2 1 ) , it holds E [ exp(µ2XY)1 ( X≥⌈C∗ s2 1 d1 ⌉ )] <α
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
math.ST 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Minimax optimal submatrix detection: Sharp non-asymptotic rates
Sharp non-asymptotic minimax rates are derived for submatrix detection in Gaussian matrices, with matching upper and lower bounds on the critical signal strength μ* for all parameter configurations.