Multimodel Deep Learning

Machine learning and data mining
Problems Classification Clustering Regression Anomaly detection AutoML Association rules Reinforcement learning Structured prediction Feature engineering Feature learning Online learning Semi-supervised learning Unsupervised learning Learning to rank Grammar induction
Supervised learning (classification • regression) Decision trees Ensembles Bagging Boosting Random forest k-NN Linear regression Naive Bayes Artificial neural networks Logistic regression Perceptron Relevance vector machine (RVM) Support vector machine (SVM)
Clustering BIRCH CURE Hierarchical k-means Expectation–maximization (EM) DBSCAN OPTICS Mean-shift
Dimensionality reduction Factor analysis CCA ICA LDA NMF PCA t-SNE
Structured prediction Graphical models Bayes net Conditional random field Hidden Markov
Anomaly detection k-NN Local outlier factor
Artificial neural networks Autoencoder Deep learning DeepDream Multilayer perceptron RNN LSTM GRU) Restricted Boltzmann machine SOM Convolutional neural network U-Net
Reinforcement learning Q-learning SARSA Temporal difference (TD)
Theory Bias-variance dilemma Computational learning theory Empirical risk minimization Occam learning PAC learning Statistical learning VC theory
Machine-learning venues NIPS ICML ML JMLR ArXiv:cs.LG
Glossary of artificial intelligence Glossary of artificial intelligence
Related articles List of datasets for machine-learning research Outline of machine learning

Multimodel Deep Learning (Ensemble Deep Learning, also Random Multimodel Deep Learning) is an ensemble learning method for classification tasks, that operates by constructing a multitude of deep learning models at training time and outputting the class that is the mode of the classes (classification) or mean prediction (regression) of the individual deep learning models. Multimodel Deep Learning corrects for Deep Learning's habit of overfitting to its training set, and also Multimodel Deep Learning is the machine learning task with the supervised learning task which uses labeled data (a classification or categorization).

History

The first algorithm for Ensemble Deep Learning was created by Li Deng and John C. Platt using ensemble deep learning for speech recognition, then developed by Qiu, Xueheng, et al. in 2017 which used ensemble Deep Learning for regression and time series forecasting. In 2018, a new ensemble, deep learning approach for classification was developed by Kowsari et al. called Random Multimodel Deep Learning (RMDL) for Classification.

Multimodel Deep Learning can solve the problem of finding the best deep learning architecture and structure while simultaneously improving robustness and accuracy through ensembles of deep learning architectures. Multimodel Deep Learning can accept as input a variety of data to include text, video, images, and symbolic data.

Random Multimodel Deep Learning (RMDL) architecture for classification. RMDL includes 3 Random models, one DNN classifier at left, one Deep CNN classifier at middle, and one Deep RNN classifier at right (each unit could be LSTM or GRU).

Application

Text and document classification

Document classification or document categorization is a problem in library science, information science and computer science. The task is to assign a document to one or more classes or categories. In recent years, many researchers have used deep learning for this problem. Ensemble Deep Learning has successfully improved this problem as a natural language processing task. A common evaluation set for text classification is the Reuters-21578 database dataset. Reuters-21578 is composed of a dataset of 11,228 newswires from Reuters, labeled over 46 topics. And another dataset is IMDB Movie reviews sentiment classification which is a dataset of 25,000 movie reviews from IMDB, labeled by sentiment (positive/negative). Reviews have been preprocessed, and each review is encoded as a sequence of word indexes (integers).

Image classification

A common evaluation set for image classification is the MNIST and CIFAR10 database datasets. MNIST is composed of handwritten digits and includes 60,000 training examples and 10,000 test examples. Multimodel Deep Learning has successfully been applied to image recognition. For MNIST dataset, Multimodel Deep Learning has improved the error rate to 0.51, 0.41, and 0.21 for 3, 9 and 15 random models respectively. The CIFAR10 dataset consists of 50,000 32x32 color training images, labeled over 10 categories, and 10,000 test images.

Face recognition

The RMDL has also successfully been used for face recognition.

Speech recognition

Deep learning systems have successfully improved the accuracy and robustness of speech recognition. Experimental results of Ensemble Deep Learning show a significant improvement in phone recognition accuracy.

Other applications

Multimodel Deep Learning can be applied to any kind of dataset such as text, image, speech, video, and face recognition datasets. Ensemble Deep Learning showed that combinations of DNNs, RNNs and CNNs with the parallel learning architecture outperform those obtained by conventional approaches.

Library and tools

• RMDL [1]

See also

• Applications of artificial intelligence
• Artificial neural networks
• Boltzmann machine
• Classification
• Comparison of deep learning software
• Compressed sensing
• Deep learning
• Document classification
• Echo state network
• Image classification
• Liquid state machine
• List of artificial intelligence projects
• List of datasets for machine learning research
• Natural language processing
• Reservoir computing
• Sparse coding

References

This article "Multimodel Deep Learning" is from Wikipedia. The list of its authors can be seen in its historical and/or the page Edithistory:Multimodel Deep Learning. Articles copied from Draft Namespace on Wikipedia could be seen on the Draft Namespace of Wikipedia and not main one.

📰 Article(s) of the same category(ies)

• Hyena Model (deep learning)
• Artivatic
• Gorilla Technology Group
• This Person Does Not Exist
• AI Constitution
• RNNSharp
• Deep Lake (Deep Learning)

---