Bidirectional Long Short-Term Memory Architecture Research Articles

Text Comment Classification and Sentiment Analysis Model Based on LSTM

This study addresses the challenge of sentiment analysis and text classification in user-generated comments, an essential task for understanding public opinion in the digital age. With the widespread adoption of social media, the sheer volume of text data has increased exponentially, thereby necessitating advanced analytical tools. The study integrates Long Short-Term Memory (LSTM) networks with attention mechanisms to develop a model adept at both classifying and analyzing the sentiment of textual comments. The research methodology commences with data cleaning, tokenization, and feature extraction, subsequently followed by the implementation of a bidirectional LSTM architecture augmented with an attention mechanism. The model underwent rigorous training and testing on a comprehensive dataset, yielding high accuracy rates in sentiment analysis and text classification. The results underscore the model’s proficiency in capturing sentiment nuances and underscore its potential for deployment in e-commerce and content recommendation systems, emphasizing the practical implications of this research for enhancing decision-making processes in businesses and among policymakers.

A novel hybrid BWO-BiLSTM-ATT framework for accurate offshore wind power prediction

Accurate wind power forecasting is pivotal in facilitating the efficient integration of renewable energy sources into existing power grids. This study proposes a novel hybrid framework, termed BWO-BiLSTM-ATT, which synergistically combines the strengths of a bidirectional long short-term memory (BiLSTM) network, a self-attention mechanism, and the Beluga Whale Optimization (BWO) algorithm. The BiLSTM architecture, with its unique ability to capture bidirectional temporal dependencies, effectively models the intricate dynamics present in wind power time series data. Integrating the self-attention mechanism further enhances the model's performance by discerning and emphasizing the most salient features within the input data. Furthermore, employing the BWO algorithm optimizes the model's hyperparameters, ensuring optimal configuration and enhancing its predictive accuracy and generalization capabilities. The proposed BWO-BiLSTM-ATT framework was rigorously evaluated using real-world data from an offshore wind farm in Yangjiang City, China. Comparative analyses were conducted against several baseline algorithms, including persistence, ARIMA, SVR, and vanilla LSTM models. The results demonstrate the superior performance of the BWO-BiLSTM-ATT framework, achieving higher R-squared and explained variance scores, coupled with lower error metrics such as MSE, RMSE, MedAE, and MAE. Statistical significance tests further corroborated the substantial performance improvements offered by the proposed framework. The combination of advanced deep learning techniques and metaheuristic optimization algorithms embedded in the BWO-BiLSTM-ATT framework provides a robust and accurate solution for wind power forecasting, enabling efficient management and seamless integration of renewable energy resources into existing power grids.

A deep learning model employing Bi-LSTM architecture to predict Martian ionosphere electron density using data from the Mars Global Surveyor mission

Planetary scientists and space agencies are highly intrigued by the Martian ionosphere due to its significant impact on upcoming Mars missions. Given its potential influence on satellite communication and navigation systems, measuring the altitude distribution of electron density (ED) values in the Martian ionosphere becomes a vital parameter to consider. The parameter of electron density in the Martian ionosphere serves as a tool to represent and predict the impacts of severe space weather events and solar activity on the Martian ionosphere. This study presents the development of a Bi-directional Long Short-Term Memory (Bi-LSTM) model, a deep machine learning algorithm to predict the Mars ionospheric ED values. To train and evaluate the model’s performance, we employ data samples from the Mars Global Surveyor (MGS) mission. For the preparation of input variables in the Bi-LSTM model, approximately 5600 ED altitude distribution of electron density profiles logged by the MGS between the years 1998 to 2005, which corresponded to the major period of the 23rd solar cycle, are taken into consideration. The performance of the proposed model in predicting both quiet and disturbed times is evaluated using the Root Mean Square Error (RMSE) and Mean Absolute Error (MAE) metrics. The significant minimum values of MAE (0.56 × 109) and RMSE (0.67 × 109) for the quiet day considered and MAE (0.27 × 109) and RMSE (0.31 × 109) for the solar flare disturbed day considered indicate the fair performance of the Bi-LSTM based prediction model. This research’s findings are vital for upcoming Mars missions, as precise predictions of ED of Mars ionosphere are crucial for reliable navigation and communication systems.

Sentiment analysis of student feedback using attention-based RNN and transformer embedding

Sentiment analysis systems aim to assess people’s opinions across various domains by collecting and categorizing feedback and reviews. In our study, researchers put forward a sentiment analysis system that leverages three distinct embedding techniques: automatic, global vectors (GloVe) for word representation, and bidirectional encoder representations from transformers (BERT). This system features an attention layer, with the best model chosen through rigorous comparisons. In developing the sentiment analysis model, we employed a hybrid dataset comprising students’ feedback and comments. This dataset comprises 3,820 comments, including 2,773 from formal evaluations and 1,047 generated by ChatGPT and prompting engineering. Our main motivation for integrating generative AI was to balance both positive and negative comments. We also explored recurrent neural network (RNN), gated recurrent unit (GRU), long short-term memory (LSTM), and bidirectional long short-term memory (Bi-LSTM), with and without pre-trained GloVe embedding. These techniques produced F-scores ranging from 67% to 69%. On the other hand, the sentiment model based on BERT, particularly its KERAS implementation, achieved higher F-scores ranging from 83% to 87%. The Bi-LSTM architecture outperformed other models and the inclusion of an attention layer further enhanced the performance, resulting in F-scores of 89% and 88% from the Bi-LSTM-BERT sentiment models, respectively.

AI Assisted Deceptive Content Analysis System Using Bi-LSTM

Abstract: In an age of widespread misinformation, creating accurate machine learning algorithms is vital for ensuring the integrity of information sharing. This project aims to develop a machine learning model capable of distinguishing between fake and true news articles. [1]The model utilizes a Bidirectional Long Short-Term Memory (Bi-LSTM) neural network architecture. Exploratory data analysis techniques are employed to gain in-sights into the characteristics and distributions within the dataset. Data visualization techniques aid in understanding patterns and relationships within the dataset. Additionally, unigram analysis is conducted to extract meaningful features from the text data. The datasets are then prepared for model training, involving preprocessing steps such as tokenization and vectorization. Fi-nally, the Bi-LSTM model is constructed, leveraging its ability to capture long-range dependencies in sequential data. The model is trained on the prepared datasets, optimized using appropri-ate techniques, and evaluated using metrics such as accuracy, precision, recall, and F1-score. The Bi-LSTM architecture offers the advantage of capturing long-range dependencies in sequential data, thereby enhancing the model’s ability to discern nuanced patterns in news articles. The primary objective of this project is to develop a robust and accurate system for automatically detecting fake news, thus playing a pivotal role in enhancing the dissemination of reliable information in the digital age.

Utilizing BiLSTM For Fine-Grained Aspect-Based Travel Recommendations Using Travel Reviews In Low Resourced Language

Recommender systems have become an essential tool for enhancing user experiences by providing personalized recommendations. In this study, we present a novel approach to constructing a recommender system specifically tailored for Malayalam travel reviews. Our objective was to extract relevant features from these reviews and employ a bidirectional Long Short-Term Memory (BiLSTM) architecture to construct a robust and accurate recommendation model. We focused on four key features extracted from the travel reviews: travel mode, travel type, location climate, and location type. The travel mode feature encompassed the mode of transport opted for the travel such as bus, car, train, etc., while the travel type captured the nature of the trip, including family, friends, or solo travel. Additionally, we considered the climate of the location, including rainy, snowy, hot, and dry, among others, and the location type, such as beach, hilly, or forest destinations. To construct our recommender system, we implemented a BiLSTM architecture, a powerful deep-learning model known for effectively capturing temporal dependencies in sequential data. This architecture allowed us to process the extracted features and learn the underlying patterns within the Malayalam travel reviews. Our experiments were conducted on a comprehensive dataset of Malayalam travel reviews, carefully curated for this study. The dataset encompassed a diverse range of travel experiences, enabling our model to learn from a wide variety of user preferences and recommendations. The performance evaluation of our recommender system yielded promising results. With an accuracy of 83.65 percent, our model showcased its ability to accurately predict and recommend travel options based on the extracted features from the reviews. The high accuracy achieved by our model underscores the effectiveness of the BiLSTM architecture in capturing the nuances of the Malayalam language and understanding the subtle preferences expressed in travel reviews. The practical implications of our work are significant, as it offers a valuable tool for travelers seeking personalized recommendations based on their travel preferences. The use of the Malayalam language in this context expands the reach of recommender systems to a wider audience, catering specifically to individuals who prefer to consume content and make decisions in their native language.

Using a Minimalist Bi-LSTM for Multi-Faceted Bearing Fault Detection

A bidirectional long short-term memory (Bi-LSTM) model is developed to predict a multi-faceted bearing health characteristic using time series vibration measurements from the Case Western Reserve University (CWRU) seeded fault test data. A maximal amount of data is applied with minimal preprocessing to test the model’s capabilities and several hyperparameters are tuned to maximize model performance. The utility of the CWRU dataset for related problems and the applicability of Bi-LSTM architecture for time series data is highlighted. The proposed model achieved a final test prediction accuracy of 98.42% and had low computation time, making it an interesting candidate for application in bearing fault prognosis.

A novel bidirectional LSTM deep learning approach for COVID-19 forecasting

COVID-19 has resulted in significant morbidity and mortality globally. We develop a model that uses data from thirty days before a fixed time point to forecast the daily number of new COVID-19 cases fourteen days later in the early stages of the pandemic. Various time-dependent factors including the number of daily confirmed cases, reproduction number, policy measures, mobility and flight numbers were collected. A deep-learning model using Bidirectional Long-Short Term Memory (Bi-LSTM) architecture was trained on data from 22nd Jan 2020 to 8 Jan 2021 to forecast the new daily number of COVID-19 cases 14 days in advance across 190 countries, from 9 to 31 Jan 2021. A second model with fewer variables but similar architecture was developed. Results were summarised by mean absolute error (MAE), root mean squared error (RMSE), mean absolute percentage error (MAPE), and total absolute percentage error and compared against results from a classical ARIMA model. Median MAE was 157 daily cases (IQR: 26–666) under the first model, and 150 (IQR: 26–716) under the second. Countries with more accurate forecasts had more daily cases and experienced more waves of COVID-19 infections. Among countries with over 10,000 cases over the prediction period, median total absolute percentage error was 33% (IQR: 18–59%) and 34% (IQR: 16–66%) for the first and second models respectively. Both models had comparable median total absolute percentage errors but lower maximum total absolute percentage errors as compared to the classical ARIMA model. A deep-learning approach using Bi-LSTM architecture and open-source data was validated on 190 countries to forecast the daily number of cases in the early stages of the COVID-19 outbreak. Fewer variables could potentially be used without impacting prediction accuracy.

Unified time series analysis with Bi-long short-term memory model for early prediction of dyslipidemia in steel workers

Dyslipidemia is a serious problem in steel workers, which needs to be detected in the early stage. However, the state of art machine learning methods failed to provide accurate classification, which resulted in reduced accuracy. So, this article aimed topredict dyslipidemia among steel workers using a recurrent neural network (RNN) with a bidirectional long short-term memory (Bi-LSTM) architecture. Initially, the data on the steel workers and their lipid profiles are collected and preprocessed by normalizing and scaling to ensure that all the features have the same scale and variance. An RNN model with a Bi-LSTM architecture is built and trained on the training set, using Adam loss functions and optimizer. The proposed RNN-BiLSTM model had an accuracy of 93.13%, specificity of 91.23%, precision of 92.23%, recall of 93.45%, and F1-score of 94.03%, which are superior to existing methods.

DRF codes: Deep SNR-robust feedback codes

We present a new Deep Neural Network (DNN)-based error correction code for fading channels with output feedback, called the Deep SNR-Robust Feedback (DRF) code. At the encoder, parity symbols are generated by a Long Short Term Memory (LSTM) network based on the message, as well as the past forward channel outputs observed by the transmitter in a noisy fashion. The decoder uses a bidirectional LSTM architecture along with a Signal to Noise Ratio (SNR)-aware attention NN to decode the message. The proposed code overcomes two major shortcomings of DNN-based codes over channels with passive output feedback: (i) the SNR-aware attention mechanism at the decoder enables reliable application of the same trained NN over a wide range of SNR values; (ii) curriculum training with batch size scheduling is used to speed up and stabilize training while improving the SNR-robustness of the resulting code. We show that the DRF codes outperform the existing DNN-based codes in terms of both the SNR-robustness and the error rate in an Additive White Gaussian Noise (AWGN) channel with noisy output feedback. In fading channels with perfect phase compensation at the receiver, DRF codes learn to efficiently exploit knowledge of the instantaneous fading amplitude (which is available to the encoder through feedback) to reduce the overhead and complexity associated with channel estimation at the decoder. Finally, we show the effectiveness of DRF codes in multicast channels with feedback, where linear feedback codes are known to be strictly suboptimal. These results show the feasibility of automatic design of new channel codes using DNN-based language models.

An Automated Model for Child Language Impairment Prediction Using Hybrid Optimal BiLSTM

Children without obvious disabilities (hearing loss/low intellectual capacity) may have language skill development issues due to specific language impairment (SLI), a communication disorder. The SLI has a significant impact on a child's speaking, listening, reading, and writing abilities. SLI is typically known as development language disorder, developmental dysphasia, or language delay. Recently, machine learning as well as deep learning techniques have been quite effective in predicting the early stage of SLI, analyzing the disorder severity, and predicting the treatment efficiency. Existing approaches primarily exploited auditory indicators to diagnose communication disorders, frequently leaving out hidden information acquired in the temporal domain. To overcome this drawback, an optimized Bidirectional Long Short Term Memory (BiLSTM) architecture is presented in this paper to handle the speech dynamics. The Improved Hybrid Aquila Optimizer and Flow Directional algorithm known as IHAOFDA is integrated with the BiLSTM architecture to optimize the hyperparameters of the BiLSTM structure. When assessed using the information from the SLI children in the Laboratory of Artificial Neural Network Applications (LANNA) dataset, the proposed model performs better. The IHAOFDA-optimized BiLSTM architecture improves accuracy in classifying different severity levels such as mild, moderate, and severe.

A Novel Hybrid Deep Learning Model for Detecting and Classifying Non-Functional Requirements of Mobile Apps Issues

As a result of the speed and availability of the Internet, mobile devices and apps are in widespread usage throughout the world. Thus, they can be seen in the hands of nearly every person, helping us in our daily activities to accomplish many tasks with less effort and without wasting time. However, many issues occur while using mobile apps, which can be considered as issues of functional or non-functional requirements (NFRs). Users can add their comments as a review on the mobile app stores that provide for technical feedback, which can be used to improve the software quality and features of the mobile apps. Minimum attention has been given to such comments by scholars in addressing, detecting, and classifying issues related to NFRs, which are still considered challenging. The purpose of this paper is to propose a hybrid deep learning model to detect and classify NFRs (according to usability, reliability, performance, and supportability) of mobile apps using natural language processing methods. The hybrid model combines three deep learning (DL) architectures: a recurrent neural network (RNN) and two long short-term memory (LSTM) models. It starts with a dataset construction extracted from the user textual reviews that contain significant information in the Arabic language. Several experiments were conducted using machine learning classifiers (MCLs) and DL, such as ANN, LSTM, and bidirectional LSTM architecture to measure the performance of the proposed hybrid deep learning model. The experimental results show that the performance of the proposed hybrid deep learning model outperforms all other models in terms of the F1 score measure, which reached 96%. This model helps mobile developers improve the quality of their apps to meet user satisfaction and expectations by detecting and classifying issues relating to NFRs.

A Deep Learning Approach for Identifying and Discriminating Spoken Arabic Among Other Languages

Spoken Language Identification (SLID) is an important step in speech-to-speech translation systems and multi-lingual automatic speech recognition. In recent research, deep learning mechanisms have been the prevailing approaches for spoken language identification. This paper aims to study, detect, and analyze spoken languages similar to Arabic in pronouncing certain words and then proposes a deep learning-based architecture, specifically the Bidirectional Long Short Term Memory (BLSTM), for spoken Arabic language identification and discrimination between these similar languages, namely, German, Spanish, French, and Russian, all of which are taken from Mozilla speech corpus languages. Additionally, our work involves a linguistic study of these considered languages. A total of ten thousand speakers are chosen for all five languages, and the BLSTM architecture is designed and implemented using acoustic signal features and applied to five experiments in this paper. The results show a precision of 98.97%, 98.73%, 98.47%, and 99.75% for identifying the spoken Arabic language separately along with German, Spanish, French, and Russian, respectively. Additionally, we achieved an average accuracy of 95.15% for discriminating between all these considered five languages in terms of the pronunciation of words. Our findings confirm that a BLSTM architecture is able to distinguish between observable similar pronunciations of words in considered languages.

Multivariate Time Series Sensor Feature Forecasting Using Deep Bidirectional LSTM

Multivariate Time series data forecasting (MTSF) is the assignment of forecasting future estimates of a particular series employing historic data. Lately, this work has enticed the focus of machine and deep learning researchers to tackle the complex and time consuming aspects of conventional forecasting techniques. Along with increasing access to historic documented data and the dire need of carrying out precise time-series feature forecasting. In this paper, we put forward a deep-learning(DL) technique proficient to tackle the setback of conventional forecasting techniques and display precise forecasting. The proposed technique is a Stacked Bidirectional long-short term memory architecture, which is an improvised version of existing Bidirectional LSTM in which multiple Bidirectional LSTM blocks are stacked, such that each layer contains multiple cells. In the direction of fair assessment, the functioning of the proposed technique is compared with existing LSTMs. Using varied evaluation measures, the obtained results show that the proposed Deep Bidirectional long-short term memory model exceeds standard approaches.

An efficient lung disease classification from X-ray images using hybrid Mask-RCNN and BiDLSTM

Lung diseases mainly affect the inner lining of the lungs causing complications in breathing, airway obstruction, and exhalation. Identifying lung diseases such as COVID-19, pneumonia, fibrosis, and tuberculosis at the earlier stage is a great challenge due to the availability of insufficient laboratory kits and image modalities. The rapid progression of the lung disease can be easily identified via Chest X-rays and this serves as a major boon for the terminally ill patients admitted to Intensive Care Units (ICU). To enhance the decision-making capability of the clinicians, a novel lung disease prediction framework is proposed using a hybrid bidirectional Long-Short-Term-Memory (BiDLSTM)-Mask Region-Based Convolutional Neural Network (Mask-RCNN) model. The Crystal algorithm is used to optimize the scalability and convergence issues in the Mask-RCNN model by hyperparameter tuning. The long-range dependencies for lung disease prediction are done using the BiDLSTM architecture which is connected to the fully connected layer of the Mask RCNN model. The efficiency of the proposed methodology is evaluated using three publicly accessible lung disease datasets namely the COVID-19 radiography dataset, Tuberculosis (TB) Chest X-ray Database, and National Institute of Health Chest X-ray Dataset which consists of the images of infected lung disease patients. The efficiency of the proposed technique is evaluated using different performance metrics such as Accuracy, Precision, Recall, F-measure, Specificity, confusion matrix, and sensitivity. The high accuracy obtained when comparing the proposed methodology with conventional techniques shows its efficiency of it in improving lung disease diagnosis.

Automatic cardiac arrhythmia classification based on hybrid 1-D CNN and Bi-LSTM model

Cardiovascular diseases (CVDs) are a group of heart and blood vessel ailments that can cause chest pain and trouble breathing, especially while active. However, some patients with heart disease have no symptoms and may benefit from screening. Electrocardiogram(ECG) measures electrical activity of the heart using sensors positioned on the skin over the chest, and it can be used for the timely detection of CVDs. This work presents a technique for classification among lethal CVDs like atrial fibrillation (Afib), ventricular fibrillation (Vfib), ventricular tachycardia (Vtec), and normal (N) beats. A novel combination of Stationary wavelet transforms (SWT) and a two-stage median filter with Savitzky–Golay(SG) filter were utilised for pre-processing of the ECG signal followed by segmentation and z-score normalisation process. Next, 1-D six-layers convolutional neural network (1-D CNN) was used for automated and reliable feature extraction. After that, bidirectional long short-term memory (Bi-LSTM) was used in the back end for classification of arrhythmias. The novelty of the present work is the use of 1-D CNN and Bi-LSTM architecture followed by relevant and effective pre-processing of the ECG signal makes this technique accurate and reliable. An accuracy of 99.41 % was achieved using 10-fold cross validation, which is superior to the existing state-of-art methods. Thus, this method presents a noble, accurate, and reliable method for classification of cardiac arrhythmia beats.

A Clinical Decision Support System for Diabetes Patients with Deep Learning: Experience of a Taiwan Medical Center.

Background: Diabetes mellitus (DM) is a major public health problem worldwide. It involves dysfunction of blood sugar regulation resulting from insulin resistance, inadequate insulin secretion, or excessive glucagon secretion.Methods: This study collated 971,401 drug usage records of 51,009 DM patients. These data include patient identification code, age, gender, outpatient visiting dates, visiting code, medication features (included items, doses, and frequencies of drugs), HbA1c results, and testing time. We apply a random forest (RF) model for feature selection and implement a regression model with the bidirectional long short-term memory (Bi-LSTM) deep learning architecture. Finally, we use the root mean square error (RMSE) as the evaluation index for the prediction model.Results: After data cleaning, the data included 8,729 male and 9,115 female cases. Metformin was the most important feature suggested by the RF model, followed by glimepiride, acarbose, pioglitazone, glibenclamide, gliclazide, repaglinide, nateglinide, sitagliptin, and vildagliptin. The model performed better with the past two seasons in the training data than with additional seasons. Further, the Bi-LSTM architecture model performed better than support vector machines (SVMs).Discussion & Conclusion: This study found that Bi-LSTM models is a well kernel in a CDSS which help physicians' decision-making, and the increasing the number of seasons will negative impact the performance. In addition, this study found that the most important drug is metformin, which is recommended as first-line treatment OHA in various situations for DM patients.

EMGHandNet: A hybrid CNN and Bi-LSTM architecture for hand activity classification using surface EMG signals

Recently, Convolutional Neural Networks (CNNs) have been used for the classification of hand activities from surface Electromyography (sEMG) signals. However, sEMG signal has spatial sparsity due to position of electrodes on hand muscle and temporal dependency due to performance of activity over a period of time. The CNN has the ability to extract spatial features and is limited in extracting temporal dependencies. Whereas, the Long Short-Term Memory (LSTM) aims to encode the temporal relations from sequential data. Hence, in this paper, we propose a hybrid CNN and Bidirectional LSTM (Bi-LSTM) based EMGHandNet architecture to encode the inter-channel and temporal dependencies of sEMG signals for hand activity classification. First, the CNN layers are used to extract deep features from sEMG signals, then these feature maps are processed by the Bi-LSTM to extract the sequential information in both the forward and backward directions. Thus, the proposed model learns both inter-channel and bidirectional temporal information in an end-to-end manner. The proposed model is trained and tested on five benchmark datasets, including the NinaPro DB1, NinaPro DB2, NinaPro DB4, BioPatRec DB2 and UCI Gesture. The average classification accuracies for the NinaPro DB1, NinaPro DB2, NinaPro DB4 and UCI Gesture are 95.77%,95.9%,91.65%, and 98.33% respectively. They correspond to an improvement of 4.42%,12.2%,18.65% and 1.33% over the respective state-of-the-art models. Moreover, for the BioPatRec DB2 dataset, a comparable performance (91.29%) is observed. The experimental results and comparisons confirm the superiority of the proposed model for hand activity classification from the sEMG signals.

Bi-LSTM Model to Recognize Human Activities in UAV Videos using Inflated I3D-ConvNet

Human activity recognition in aerial videos is an emerging research area. In this paper, an Inflated I3D-ConvNet (Inflated I3D) and Bidirectional Long Short-Term Memory (Bi-LSTM) based human action recognition model in UAV videos have been proposed. The initial module was pre-trained using the Kinetics-400 video dataset, which consisted of 400 classes of human activities and around 400 video clips for each class culled from real-world and arduous YouTube videos. The proposed inflated I3D-ConvNet which was built on 2D-ConvNet inflation learns and extracts spatio-temporal features from aerial video while leveraging the architectural design of Inception-V1. The proposed model employs Bi-LSTM architecture for human action classification on the Drone-Action dataset which is a smaller benchmark UAV-captured video dataset. This model considerably improves the state-of-the-art results in activity classification using the SoftMax classifier and retains an accuracy of about 98.4%.

Radar Signals Intrapulse Modulation Recognition Using Phase-Based STFT and BiLSTM

The goal of Radar Emitter Recognition (RER) is to extract the features of a received emitter signal. It has become a critical issue as new radar types are emerging and the electromagnetic environment is becoming more dense and complex. Deep Neural Networks (DNNs) have recently proven effective for emitter identification, however, recognition of phase-coded waveforms at a low Signal to Noise Ratio (SNR) remains a challenge. In this paper, a novel phase-based RER approach using Short Time Fourier Transform (STFT) and Bidirectional Long Short Term Memory (BiLSTM) is proposed while enhancing ability of learning features from noisy signals. The phase spectrum of phase-coded signals is analyzed in contrast to the amplitude spectrum used in the state-of-the-art approaches in the literature. The derived phase-based features are directly provided as an input to the proposed BiLSTM architecture. A fully connected layer follows the BiLSTM layer. Finally, a softmax classifier is employed to accomplish the recognition task. Six distinct types of phase-coded waveforms degraded by Additive White Gaussian Noise (AWGN) with SNRs ranging from -8dB to 8dB are simulated. The suggested method in this research involves simple pre-processing and exhibits overall recognition accuracy of more than 90% at SNR of -2dB.