Measuring the accuracy of LSTM and BiLSTM models in the application of artificial intelligence by applying chatbot programme

Abstract

Python programme contains a question and answer system that derived from data sets that have used and implemented the chatbot in this modern era. where the data collected is in the form of corpuses containing extensive metadata-rich fictional conversations derived from extracted film scripts, commonly called cornell movie dialogue corpus. The various models have been used chatbots in python programmes, and LSTM and BiLSTM models were specifically used in this study. Where the form of accuracy will be reported as a result of the implementation of LSTM and BiLSTM models in the chatbot programme. The programme performance will be influenced by the data from the model selection, because the level of accuracy is determined by the target programme being taken. So this is the main factor that determines which model to choose. Based on considerations required for choosing the programme model, in the end the LSTM and the BiLSTM models are chosen and will be applied to the programme. Based on the LSTM and BiLSTM chatbot programmes that have been tested, it can be concluded that the best parameters come from a pair of BiLSTM chatbots using the BiLTSM model with an average accuracy value of 0.995217.

Deep learning in phishing mitigation: a uniform resource locator-based predictive model

“…In addition, fine-tuning neurons in the running layers could significantly affect deep learning prediction accuracy and minimize the elapsed time (time consumption by training and testing pipelines). Hence, the dual deep learning predictive model like that of CNN-BiLSTM architecture could emulate the training imbalanced cyber-data stream into an informative feature space without compromising its effectiveness and efficiency in a sandboxing context [42], [43]. According to the results shown in Table 5, our technique offers comparable accuracy, recall, and precision to the current state-of-the-art models.…”

“…Contrarily, a solo deep learning predictive model (not a cross-combination or blended) the training pipeline and testing pipeline along with the number of epochs. Thus, the best improvement to margin the prediction setting between under-fitting and overfitting can be obtained by the early stop of running epochs (no more than ten via CNN-BiLSTM) [16], [27], [42]. In addition, fine-tuning neurons in the running layers could significantly affect deep learning prediction accuracy and minimize the elapsed time (time consumption by training and testing pipelines).…”

Comparative Analysis of Performance between Multimodal Implementation of Chatbot Based on News Classification Data Using Categories

“…Here, Cornell Movie Dialog Corpus is used by the author, where it contains a dataset in the form of a corpus, which consists of a metadata-rich fictional collection extracted from movie scripts. Based on the experience of the dataset, various other types of models will be tested in this study [38]. In this research experimental setup, the proportion of detailed train data is 144,000, which is 80 percent of the total dataset, and the test data is 36,000, which is 20 percent of the total dataset.…”

Deep Feature Vectors Concatenation for Eye Disease Detection Using Fundus Image

“…The method we propose in this study, namely, the concatenate model of two CNN models (DenseNet121 and Inception-ResNetV2), outperformed those of previous studies, with an accuracy of 91%. In the future, this model should be improved and evaluated on image datasets of much better quality and quantity to recognise the five severity levels of DR, thus resulting in a better research [46].…”