Selectively Ensembling Neural Classifiers
نویسندگان
چکیده
Ensembling neural classifiers can significantly improve the generalization ability of classification systems. In this paper, GASEN, a genetic algorithm based selective ensemble method that has been shown to be excellent in ensembling neural regressors, is applied to neural classifiers. Experiments on four large data sets show that this method can generate ensembles of neural classifiers with stronger generalization ability than those generated by Bagging, Adaboost, or Arc-x4.
منابع مشابه
Document Image Classification with Intra-Domain Transfer Learning and Stacked Generalization of Deep Convolutional Neural Networks
In this work, a region-based Deep Convolutional Neural Network framework is proposed for document structure learning. The contribution of this work involves efficient training of region based classifiers and effective ensembling for document image classification. A primary level of ‘inter-domain’ transfer learning is used by exporting weights from a pre-trained VGG16 architecture on the ImageNe...
متن کاملCS 224D Final Project: Neural Network Ensembles for Sentiment Classification
We investigate the effect of ensembling on two simple models: LSTM and bidirectional LSTM. These models are used for fine-grained sentiment classification on the Stanford Sentiment Treebank dataset. We observe that ensembling improves the classification accuracy by about 3% over single models. Moreover, the more complex model, bidirectional LSTM, benefits more from ensembling.
متن کاملEnsemble Neural Relation Extraction with Adaptive Boosting
Relation extraction has been widely studied to extract new relational facts from open corpus. Previous relation extraction methods are faced with the problem of wrong labels and noisy data, which substantially decrease the performance of the model. In this paper, we propose an ensemble neural network model Adaptive Boosting LSTMs with Attention, to more effectively perform relation extraction. ...
متن کاملDenResNet: Ensembling Dense Networks and Residual Networks
We combine various state of the art approaches to training deep convolutional neural networks to achieve the best performance possible on the Tiny ImageNet dataset. We emphasize the depth of the network through residual networks, transfer learning with pretrained models, and ensemble methods. We achieved a final ensemble test error of 25.6%, which places us at the top of the leaderboard. We fur...
متن کاملEnsemble Pruning of Internet Traffic Classifiers for Security Applications
Internet Traffic classification is vital for various network activities such as detection of malware. Security is major issue of concern which accounts for the reputation and reliability of system. Malware effects the system adversely results in data loss or abnormal functioning. Hence, detection and as well as removal of malware is crucial. Combining set of classifiers called as ensembling pro...
متن کامل