Repository contain different neural network implementation:

1) sentiment analysis simple - simple implementation of regular neural network (NN) with four fully-connected layers for 
                               classification of product reviews - task that was considered in Yandex_sentiment_analysis repository.

2) sentiment analysis moment adaptive learnrate - implementation of regular NN using special technics like introduction momentum and
                                                  adaptive learnrates (RMSProp, Adam optimizer, dropout)

3) sentiment analysis tensorflow keras - this time NN was written using keras and tensorflow modules

4) convolution NN - convolution NN for face emotion classification

5) autoencoders - unsupervised machine learning - autoencoders, Restricted Bolzman machines, convolutional autoencoders; 
                  deep neural and deep belief networks

6) recurrent neural networks -