Multilabel Image classification using optimized ensemble deep learning

Deep learning is presently giving very good performance for many applications like Image identification, Speech recognition, Natural language processing, Recommendation systems etc. Ensemble learning is giving very good performance when applied in deep learning. In this paper, we use deep learning as well as ensemble learning for the classification of multi-label dataset. The novelty of the model is that we combine three different models. In the first model we use AutoEncoder+CNN by using only the encoder part to reduce the dimension. The second model is created by adding the noise to the input data and then using AutoEncoder to denoise the noisy data so that the model can work even for noisy dataset in real-time applications. The third model uses the vision Transformer. The proposed ensemble deep learning model combines these different models to predict the final result. The ensemble model outperforms each individual model and the state-of-the-art models.