Ensemble of Networks for Multilabel Classification

Multilabel learning goes beyond standard supervised models by associating a sample with more than one class label. Among the many techniques developed in last decade to handle multilabel best approaches are those harnessing power of ensembles and deep learners. This work proposes merging both methods combining set gated recurrent units, temporal convolutional neural networks, long short-term memory networks trained variants Adam optimization approach. We examine variants, each fundamentally based on difference between present past gradients, step size adjusted for parameter. also combine Incorporating Multiple Clustering Centers bootstrap-aggregated decision trees ensemble, which is shown further boost classification performance. In addition, we provide an ablation study assessing performance improvement that module our ensemble produces. experiments large datasets representing wide variety tasks demonstrate robustness outperform state-of-the-art.