Novel approach of encrypted network traffic classification using deep convolutional neural network with Artificial Bee Colony and Genetic Algorithm

A CNN-ABC model for estimation and optimization of heat generation rate and voltage distributions of lithium-ion batteries for electric vehicles

Artificial bee colony (ABC) algorithm has proved its utilization in solving various problems including engineering optimization problems. ABC algorithm is most popular and youngest member of the family of population based nature inspired meta-heuristic swarm intelligence method. ABC has been proved its superiority over some other Nature Inspired Algorithms (NIA) when applied for both benchmark functions and real world problems. The performance of search process of ABC depends on a random value which tries to balance exploration and exploitation phase. In order to increase the performance it is required to balance the exploration of search space and exploitation of optimal solution of the ABC. This paper suggests a modified ABC algorithm. The proposed method integrates crossover operation from Genetic Algorithm (GA) with original ABC algorithm. The proposed method is named as Crossover based ABC (CB-ABC). The CB-ABC strengthens the exploitation phase of ABC as crossover enhances exploration of search space. CB-ABC tested over some standard benchmark functions and a well known continuous optimization problems.

Action recognition is a challenging task. Deep learning models have been investigated to solve this problem. Setting up a new neural network model is a crucial and time-consuming process. Alternatively, pre-trained convolutional neural network (CNN) models offer rapid modeling. The selection of the hyperparameters of CNNs is a challenging issue that heavily depends on user experience. The parameters of CNNs should be carefully selected to get effective results. For this purpose, the artificial bee colony (ABC) algorithm is used for tuning the parameters to get optimum results. The proposed method includes three main stages: the image preprocessing stage involves automatic cropping of the meaningful area within the images in the data set, the transfer learning stage includes experiments with six different pre-trained CNN models and the hyperparameter tuning stage using the ABC algorithm. Performance comparison of the pre-trained CNN models involving the use and nonuse of the ABC algorithm for the Stanford 40 data set is presented. The experiments show that the pre-trained CNN models with ABC are more successful than pre-trained CNN models without ABC. Additionally, to the best of our knowledge, the improved NASNet-Large CNN model with the ABC algorithm gives the best accuracy of 87.78% for the overall success rate-based performance metric.

In this paper, the primary focus is of Slot Tagging of Gujarat Dialogue, which enables the Gujarati language communication between human and machine, allowing machines to perform given task and provide desired output. The accuracy of tagging entirely depends on bifurcation of slots and word embedding. It is also very challenging for a researcher to do proper slot tagging as dialogue and speech differs from human to human, which makes the slot tagging methodology more complex. Various deep learning models are available for slot tagging for the researchers, however, in the instant paper it mainly focuses on Long Short-Term Memory (LSTM), Convolutional Neural Network - Long Short-Term Memory (CNN-LSTM) and Long Short-Term Memory – Conditional Random Field (LSTM-CRF), Bidirectional Long Short-Term Memory (BiLSTM), Convolutional Neural Network - Bidirectional Long Short-Term Memory (CNN-BiLSTM) and Bidirectional Long Short-Term Memory – Conditional Random Field (BiLSTM-CRF). While comparing the above models with each other, it is observed that BiLSTM models performs better than LSTM models by a variation ~2% of its F1-measure, as it contains an additional layer which formulates the word string to traverse from backward to forward. Within BiLSTM models, BiLSTM-CRF has outperformed other two Bi-LSTM models. Its F1-measure is better than CNN-BiLSTM by 1.2% and BiLSTM by 2.4%.KeywordsSpoken Language Understanding (SLU)Long Short-Term Memory (LSTM)Slot taggingBidirectional Long Short-Term Memory (BiLSTM)Convolutional Neural Network - Bidirectional Long Short-Term Memory (CNN-BiLSTM)Bidirectional Long Short-Term Memory (BiLSTM-CRF)

The aim of this book is to present and analyse theoretical advances and also emerging practical applications of swarm and evolutionary intelligence. It comprises nine chapters. Chapter 1 provides a theoretical introduction of the computational optimization techniques regarding the gradient-based methods such as steepest descent, conjugate gradient, newton and quasi-Newton methods and also the non-gradient methods such as genetic algorithm and swarm intelligence algorithms. Chapter 2, discusses evolutionary computation techniques and genetic algorithm. Swarm intelligence theory and particle swarm optimization algorithm are reviewed in Chapter 3. Also, several variations of particle swarm optimization algorithm are analysed and explained such as Geometric PSO, PSO with mutation, Chaotic PSO with mutation, multi-objective PSO and Quantum mechanics – based PSO algorithm. Chapter 4 deals with two essential colony bio-inspired algorithms: Ant colony optimization (ACO) and Artificial bee colony (ABC). Chapter 5, presents and analyses Cuckoo search and Bat swarm algorithms and their latest variations. In chapter 6, several other metaheuristic algorithms are discussed such as: Firefly algorithm (FA), Harmony search (HS), Cat swarm optimization (CSO) and their improved algorithm modifications. The latest Bio-Inspired Swarm Algorithms are discussed in chapter 7, such as: Grey Wolf Optimization (GWO) Algorithm, Whale Optimization Algorithm (WOA), Grasshopper Optimization Algorithm (GOA) and other algorithm variations such as binary and chaotic versions. Chapter 8 presents machine learning applications of swarm and evolutionary algorithms. Illustrative real-world examples are presented with real datasets regarding neural network optimization and feature selection, using: genetic algorithm, Geometric PSO, Chaotic Harmony Search, Chaotic Cuckoo Search, and Evolutionary Algorithm and also crime forecasting using swarm optimized SVM. In chapter 9, applications of swarm intelligence on deep long short-term memory (LSTM) networks and Deep Convolutional Neural Networks (CNNs) are discussed, including LSTM hyperparameter tuning and Covid19 diagnosis from chest X-Ray images. The aim of the book is to present and discuss several state-of-theart swarm intelligence and evolutionary algorithms together with their variances and also several illustrative applications on machine learning and deep learning.

A genetic Artificial Bee Colony algorithm for signal reconstruction based big data optimization

Evolutionary algorithms (EAs) have been widely used in recent years. Artificial bee colony (ABC) algorithm is an EA for numerical optimisation problems. Recently, more and more researchers show interest in ABC algorithm. Previous studies have shown that the ABC algorithm is an efficient, effective and robust evolutionary optimisation method. However, the convergence rate of ABC algorithm still does not meet our requirements and it is necessary to optimise the ABC algorithm. In this paper, several local search operations are embedded into the ABC algorithm. This modification enables the algorithm to get a better balance between the convergence rate and the robustness. Thus it can be possible to increase the convergence speed of the ABC algorithm and thereby obtain an acceptable solution. Such an improvement can be advantageous in many real-world problems. This paper focuses on the performance of improving artificial bee colony algorithm with differential strategy on the numerical optimisation problems. The proposed algorithm has been tested on 18 benchmark functions from relevant literature. The experiment results indicated that the performance of the improved ABC algorithm is better than that of the original ABC algorithm and some other classical algorithms.

Artificial bee colony (ABC) algorithm, that mimics the intelligent foraging behavior of honey bees, is increasingly gaining acceptance in the field of process optimization, as it is capable of handling nonlinearity, complexity and uncertainty. Straight cylindrical turning is a complex and nonlinear machining process which involves the selection of appropriate cutting parameters that affect the quality of the workpiece. This paper presents the estimation of optimal cutting parameters of the straight cylindrical turning process using the ABC algorithm. The ABC algorithm is first tested on four benchmark problems of numerical optimization and its performance is compared with genetic algorithm (GA) and ant colony optimization (ACO) algorithm. Results indicate that, the rate of convergence of ABC algorithm is better than GA and ACO. Then, the ABC algorithm is used to predict optimal cutting parameters such as cutting speed, feed rate, depth of cut and tool nose radius to achieve good surface finish. Results indicate that, the ABC algorithm estimated a comparable surface finish when compared with real coded genetic algorithm and differential evolution algorithm.

In natural language processing (NLP), text classification is a crucial task that involves analyzing textual data, which often has high dimensionality. A good word representation is essential to address this challenge, and the word representation using GloVe is one of the popular methods that provides pre-trained word representations in high-dimensional vectors. This research evaluates the effectiveness of three deep learning techniques Convolutional Neural Network (CNN), Deep Neural Network (DNN), and Long Short-Term Memory (LSTM) for online news classification using 300-dimensional GloVe word representations. The CNN model utilizes convolutional and pooling layers to extract local features, the DNN relies on dense layers to learn abstract representations, while the LSTM excels at capturing long-term dependencies between words. The results show that the LSTM model achieved the best accuracy at 93.45%, followed by CNN at 91.24%, and DNN at 90.67%. The superiority of LSTM is attributed to its ability to effectively capture temporal relationships and context, while CNN offers efficiency with faster training times. Although DNN produced solid performance, it is less optimal in understanding word sequences. These findings indicate that LSTM outperforms the other models in online news text classification tasks.

Answer selection (AS) is a critical subtask of the open-domain question answering (QA) problem. The present paper proposes a method called RLAS-BIABC for AS, which is established on attention mechanism-based long short-term memory (LSTM) and the bidirectional encoder representations from transformers (BERT) word embedding, enriched by an improved artificial bee colony (ABC) algorithm for pretraining and a reinforcement learning-based algorithm for training backpropagation (BP) algorithm. BERT can be comprised in downstream work and fine-tuned as a united task-specific architecture, and the pretrained BERT model can grab different linguistic effects. Existing algorithms typically train the AS model with positive-negative pairs for a two-class classifier. A positive pair contains a question and a genuine answer, while a negative one includes a question and a fake answer. The output should be one for positive and zero for negative pairs. Typically, negative pairs are more than positive, leading to an imbalanced classification that drastically reduces system performance. To deal with it, we define classification as a sequential decision-making process in which the agent takes a sample at each step and classifies it. For each classification operation, the agent receives a reward, in which the prize of the majority class is less than the reward of the minority class. Ultimately, the agent finds the optimal value for the policy weights. We initialize the policy weights with the improved ABC algorithm. The initial value technique can prevent problems such as getting stuck in the local optimum. Although ABC serves well in most tasks, there is still a weakness in the ABC algorithm that disregards the fitness of related pairs of individuals in discovering a neighboring food source position. Therefore, this paper also proposes a mutual learning technique that modifies the produced candidate food source with the higher fitness between two individuals selected by a mutual learning factor. We tested our model on three datasets, LegalQA, TrecQA, and WikiQA, and the results show that RLAS-BIABC can be recognized as a state-of-the-art method.

The digital age has added a new term to the literature of information and computer sciences called as the big data in recent years. Because of the individual properties of the newly introduced term, the definitions of the data-intensive problems including optimization problems have been substantially changed and investigations about the solving capabilities of the existing techniques and then developing their specialized variants for big data optimizations have become important research topic. Artificial Bee Colony (ABC) algorithm inspired by the clever foraging characteristics of the real honey bees is one of the most successful swarm intelligence-based metaheuristics. In this study, a new ABC algorithm-based technique that is named source-linked ABC (slinkABC) was proposed by considering the properties of the optimization problems related with the big data. The slinkABC algorithm was tested on the big data optimization problems presented at the Congress on Evolutionary Computation (CEC) 2015 Big Data Optimization Competition. The results obtained from the experimental studies were compared with the different variants of the ABC algorithm including gbest-guided ABC (GABC), ABC/best/1, ABC/best/2, crossover ABC (CABC), converge-onlookers ABC (COABC), quick ABC (qABC) and modified gbest-guided ABC (MGABC) algorithms. In addition to these, the results of the proposed ABC algorithm were also compared with the results of the Differential Evolution (DE) algorithm, Genetic algorithm (GA), Firefly algorithm (FA), Phase-Based Optimization (PBO) algorithm and Particle Swarm Optimization (PSO) algorithm-based approaches. From the experimental studies, it was understood that the ABC algorithm modified by considering the unique properties of the big data optimization problems as in the slinkABC produces better solutions for most of the tested instances compared to the mentioned optimization techniques.

Coronavirus (COVID-19) is an infectious disease that has spread across the world within a short period of time and is causing rapid casualties. The main symptoms of this virus are shortness of breath, fever, cough, and a sore throat. The virus is detected through samples, such as throat swabs and sputum, taken from people who meet the possible case definition and the results are usually obtained within a few hours or a day. The development of test kits to detect the COVID-19 virus is still an open research topic, and automated and faster diagnostic tools are needed. Recent studies have shown that biomedical images can be used for COVID-19 testing. This study proposes the hybrid use of pre-trained deep networks and the long short-term memory (LSTM) for the classification of COVID-19 from contrast-enhanced chest X-rays. In the proposed system, a transformation function is applied to X-ray images first. Then, the artificial bee colony (ABC) algorithm is used to optimize the parameters obtained from the transformation function. The pre-trained deep network models and LSTM are preferred to extract features from the contrast-enhanced chest X-rays. At the final stage, COVID-19, normal (healthy), and pneumonia chest X-ray are classified using softmax. To evaluate the performance of the proposed method, the "COVID-19 radiography" dataset, which is widely used in the literature, is preferred. From the proposed model, 98.97% accuracy, 98.80% precision, and 98.70% sensitivity rates are obtained. Experimental results reveal that the proposed model provides efficient results compared to other methods. Thanks to the application of ABC-based image enhancement, increased classification of 2.5% has been achieved against other state-of-the-art models.

Gradient boosted regression model for the degradation analysis of prismatic cells

Maxout neurons for deep convolutional and LSTM neural networks in speech recognition

Both Convolutional Neural Networks (CNNs) and Long Short-Term Memory (LSTM) have shown improvements over Deep Neural Networks (DNNs) across a wide variety of speech recognition tasks. CNNs, LSTMs and DNNs are complementary in their modeling capabilities, as CNNs are good at reducing frequency variations, LSTMs are good at temporal modeling, and DNNs are appropriate for mapping features to a more separable space. In this paper, we take advantage of the complementarity of CNNs, LSTMs and DNNs by combining them into one unified architecture. We explore the proposed architecture, which we call CLDNN, on a variety of large vocabulary tasks, varying from 200 to 2,000 hours. We find that the CLDNN provides a 4-6% relative improvement in WER over an LSTM, the strongest of the three individual models.