Patrik Lauha
1 indexed paper
Recent (6 mo)
1With code
0Influential cites
0Benchmarked
0Publications per year
126
Top categories
Audio and Speech Processing×1Sound×1
Frequent co-authors
Research Timeline
2026
Mixture-Constrained Max Pooling Improves Separation-Based Bird Species Classification
This paper proposes an ensemble of two source separators, FTRNN and TF-Locoformer, trained with mixture invariant training (MixIT), and introduces mixture-constrained max pooling (MCM) to improve bird species classification from field recordings.
Highlighted terms show continued research focus across papers
Papers
eess.AScs.SDEmpiricalRecentJul 3, 2026
Mixture-Constrained Max Pooling Improves Separation-Based Bird Species Classification
Yuzhu Wang, Kalle Lahtinen, Patrik Lauha, Shiqi Zhang +3 more
This paper proposes an ensemble of two source separators, FTRNN and TF-Locoformer, trained with mixture invariant training (MixIT), and introduces mixture-constrained max pooling (MCM) to improve bird…
View →