this increase the translation performance. Here the
1this paper describes aboutneural machine translation which is new approach for translating. it creates aneural network which increase the translation performance. Here the neuralnetwork model uses encoder -decoder architecture, which encode the sourcesentence and decode it to destination sentence by predicting the target words.With this approach machine translation can be achieved.2DNN is powerful model used inmany field it is suitable for performing difficult learning tasks.in this paperwe discuss about seq2seq approach including LSTM to map the input sequence to avector fixed dimension.
it uses encoder-decoder RNN which encode the sourcesentence and decode the sentence to provide target words.3Conservation model is important task in machine learningas it understands the question and reply for thosequestion. In this paper theconservation model is built by using sequence to sequence framework. Repliesare generated by training large datasets. conservation model can be open domainor closed domain. the conservation model can vary based on the datasets exampleIT helpdesk datasets can be used to provide solutions for technical problem.
4Here open domain conservation system is build based onlarge dialogue corpora using generative model.it produces system response byunderstanding the questions with help of rnn encoder- decoder. this paper discussesabout the limitation of this approaches and show how the maximize theperformance.5In some cases seq2seq neural networks can produceresponses which does not suites to the responses so usage of Maximum Mutualinformation is increased for producing more interesting and more appropriateresponses.
6This paper proposes that in seq2seq neural networksattention and intention plays an important role. so,in order to focus onintention neural network should consist three recurrent networks. The encoderlayer for word level model represent score side sentence.
The intention networklayer for modeling the dynamics of the intension process. The decoder layer forproduce the response to the input. Neural network improves the efficiency ofthe output.7 Here persona-based model is created in order to reducehuman work. it replaces the human and provide human like responses it capturesindividual characteristics such as background info and speaking style.8In seq2seq learning (copying) i.
e. certain contents areselectively repeated in the output sequence. A similar way in humans, replicatecertain content. here we incorporate copying in the neural network and build anew model called Copynet, involved in encoder and in decoder. TheCopynet isdesigned in such a way that it generates a response by copying mechanism anduse them in proper place in the output sequence.9This paper describes about the chat- bot used in twittersocial network for entertainment and for advertisement. these bots use twitterdatasets and provide a realistic conversation like human10this paper describe application that incorporatehumanappearanceand simulate human dialog.the knowledge of bot is stored in a database by the developers.
this is just ainitial stage for AI bot start to interact with humans later bots can be builtin human appearance and are used for substitute for men.11this paper discusses about the understanding and analysesthe intelligence of chatbot. The analysis is done in order to check the intelligenceof the chatbot. There are various parameters to consider like textcategorization, entity extraction, frequency analysis and model the vocabularyusing word to vector system. Here this analysis provide metric called botintelligence score to evaluate.12 This paper proposes a hybrid neural network model whichcontains of some important network model. here the datasets are trained andexperimented results show that the best accuracy belong to different hybridmodel.
13here the human machine interface demand for a cleancommunication that are applied on various tasks. here we develop an intelligentchat machine which generate conservation sentence using RNN and single neuralnetwork model that process a conversation sentence by connecting the words.