An efficient framework for fuzzy classifiers – We propose a robust approach for fuzzy classifiers with a limited number of instances that learns to reason about (i.e. the classifiers). The proposed approach consists of three steps. First, we consider each instance separately and make the optimal decision for each instance. Second, we consider the probability of each instance to be a fuzzy probability vector, and perform a Bayesian search to identify the most informative fuzzy classifier. Finally, we perform two Bayesian optimization steps, one for each instance. The final optimization step is designed so that the decision is not made when the most informative fuzzy classifier does not exist.

This paper presents a new concept called logistic regression analysis for deep learning in neural networks (neurons) that integrates logistic regression with a deep neural network (DNN). By using a mixture of DNN models, we show that deep neural networks with logistic regression have a better performance due to the use of the logistic regression. We test several datasets of deep neural networks and use the proposed logistic regression analysis to develop a simple neural net with a DNN model which is capable of learning logistic regression on the data. Experiments on the MNIST dataset are conducted using data from MNIST 2014 dataset, MNIST 2015 dataset and MNIST 2016 dataset. The proposed logistic regression analysis also helps with the model learning on the MNIST dataset by leveraging the logistic regression analysis for training the DNN network.

Identifying and Ranking Images from Streaming Images

A new scoring approach based on Bayesian network of vowel sounds

# An efficient framework for fuzzy classifiers

Bayesian Network Subspace Revisited: A Bayesian Network Approach

An Experimental Comparison of Bayes-Encoded Loss Functions for Machine Learning with Log-Gabor FiltersThis paper presents a new concept called logistic regression analysis for deep learning in neural networks (neurons) that integrates logistic regression with a deep neural network (DNN). By using a mixture of DNN models, we show that deep neural networks with logistic regression have a better performance due to the use of the logistic regression. We test several datasets of deep neural networks and use the proposed logistic regression analysis to develop a simple neural net with a DNN model which is capable of learning logistic regression on the data. Experiments on the MNIST dataset are conducted using data from MNIST 2014 dataset, MNIST 2015 dataset and MNIST 2016 dataset. The proposed logistic regression analysis also helps with the model learning on the MNIST dataset by leveraging the logistic regression analysis for training the DNN network.