COVID-19 cases prediction by using hybrid machine learning and beetle antennae search approach

Artificial intelligence (AI) has been applied widely in our daily lives in a variety of ways with numerous successful stories. AI has also contributed to dealing with the coronavirus disease (COVID-19) pandemic, which has been happening around the globe. This paper presents a survey of AI methods being used in various applications in the fight against the COVID-19 outbreak and outlines the crucial roles of AI research in this unprecedented battle. We touch on a number of areas where AI plays as an essential component, from medical image processing, data analytics, text mining and natural language processing, the Internet of Things, to computational biology and medicine. A summary of COVID-19 related data sources that are available for research purposes is also presented. Research directions on exploring the potentials of AI and enhancing its capabilities and power in the battle are thoroughly discussed. We highlight 13 groups of problems related to the COVID-19 pandemic and point out promising AI methods and tools that can be used to solve those problems. It is envisaged that this study will provide AI researchers and the wider community an overview of the current status of AI applications and motivate researchers in harnessing AI potentials in the fight against COVID-19.

/pdf/artificial-intelligence-in-the-battle-against-coronavirus-3ayvkhps4h.pdf

Artificial Intelligence in the Battle against Coronavirus (COVID-19): A Survey and Future Research Directions

A–Z survey of Internet of Things: Architectures, protocols, applications, recent advances, future directions and recommendations

Computer vision is one of the most frontier technologies in computer science. It is used to build artificial systems to extract valuable information from images and has a broad range of applications in various areas such as agriculture, business, and healthcare. Convolutional neural networks represent the key algorithms in computer vision, and in recent years, they have attained notable advances in many real-world problems. The accuracy of the network for a particular task profoundly relies on the hyperparameters’ configuration. Obtaining the right set of hyperparameters is a time-consuming process and requires expertise. To approach this concern, we propose an automatic method for hyperparameters’ optimization and structure design by implementing enhanced metaheuristic algorithms. The aim of this paper is twofold. First, we propose enhanced versions of the tree growth and firefly algorithms that improve the original implementations. Second, we adopt the proposed enhanced algorithms for hyperparameters’ optimization. First, the modified metaheuristics are evaluated on standard unconstrained benchmark functions and compared to the original algorithms. Afterward, the improved algorithms are employed for the network design. The experiments are carried out on the famous image classification benchmark dataset, the MNIST dataset, and comparative analysis with other outstanding approaches that were tested on the same problem is conducted. The experimental results show that both proposed improved methods establish higher performance than the other existing techniques in terms of classification accuracy and the use of computational resources.

Optimizing Convolutional Neural Network Hyperparameters by Enhanced Swarm Intelligence Metaheuristics

Cloud computing is an emerging computer technology, that provides distributed, scalable, elastic computer resources to the end-user over the Internet. One of the most challenging tasks in the cloud computing environment is task scheduling. The main objectives of the task scheduling are to identify the appropriate resources for scheduling a specific task on time, utilize the resources more efficiently, and reduce the total completion time of all input tasks to be executed. The task scheduling problem belongs to the class NP-hard. Since metaheuristic algorithms are proven to be efficient in the NP hard optimization, in this paper, we propose a task scheduling algorithm using metaheuristics approach. The proposed scheduler is based on the grey wolf optimizer nature-inspired algorithm. The experimental results prove the quality and robustness of the proposed method.

Task Scheduling in Cloud Computing Environment by Grey Wolf Optimizer

The most challenging task in the machine learning domain is optimizing the hyperparameters in convolutional neural networks. This task is representative of NP-hard problems, and consequently, it is not possible to solve it by applying standard deterministic approaches in an acceptable amount of time. Additionally, convolutional neural networks’ hyperparameters must be optimized for each particular problem, as there is no solution that fits all possible applications. Swarm intelligence metaheuristics have been established as efficient optimizers, and this paper proposes the enhanced sine cosine algorithm to address the task of hyperparameters optimization. The experiments conducted in this research were executed with the CIFAR-10 benchmark dataset. The experimental results were analyzed and validated against other proven metaheuristics approaches, and it can be concluded that the presented enhanced sine cosine approach outperformed other methods included in this research.

Convolutional Neural Networks Hyperparameters Optimization Using Sine Cosine Algorithm

Abstract. Recent technological development in the computer vision field has provided opportunities for agriculture applications. The field of computer vision allows a machine to “see,” providing an alternative to using the human eye to identify, measure, and track a target in image processing. Computer vision technology uses a sensor, camera, and computer. The carrot is one of the most important vegetables in the world, and a grading scale for carrots will improve market competitiveness. In marketing and carrot processing, carrot grading plays a major part. Existing approaches using the traditional carrot grading system, which requires manual involvement, are inefficient and labor-intensive. The computer vision feature parameters of the carrot, such as its length, average diameter, maximum diameter, aspect ratio, perimeter, and area, are extracted, along with RGB (Red, Green, Blue) and HSV (Hue, Saturation, Value) color parameters. These 12 computer vision parameters are given as input to a recognition model using the proposed optimized cascaded graph convolutional neural network with Bayesian optimization. The statistical results of the experiments performed with this model show that the proposed recognition system with computer vision feature parameters can grade carrots accurately.

Carrot grading system using computer vision feature parameters and a cascaded graph convolutional neural network

One of the primary goals in designing a smart home environment is to minimize energy consumption by efficient management of home appliances. Smart homes often employ wireless sensor networks (WSN) to monitor physical parameters (like light, temperature, or presence of users at home). There is a growing concern about protection of data acquired by easily accessible wireless sensors; on the other hand, smart home networking trend stresses the importance of further protecting the communication of acquired data. Public-key cryptography methods have been long considered as energy inefficient to be applied in constraint environments such as sensor networks. In this paper we revisit public-key data protection in WSN and design a fault tolerant security architecture based on the use of elliptic curve cryptography (ECC), whose primary goal is to maximize lifetime of sensor nodes. We propose the use of Java enabled, heterogenous and clustered WSN platform and discuss its advantages over existing solutions, which commonly imply heavy optimization of security code and procedures.

Energy efficient security architecture for wireless sensor networks

As solar energy generation has become more and more important for the economies of numerous countries in the last couple of decades, it is highly important to build accurate models for forecasting the amount of green energy that will be produced. Numerous recurrent deep learning approaches, mainly based on long short-term memory (LSTM), are proposed for dealing with such problems, but the most accurate models may differ from one test case to another with respect to architecture and hyperparameters. In the current study, the use of an LSTM and a bidirectional LSTM (BiLSTM) is proposed for dealing with a data collection that, besides the time series values denoting the solar energy generation, also comprises corresponding information about the weather. The proposed research additionally endows the models with hyperparameter tuning by means of an enhanced version of a recently proposed metaheuristic, the reptile search algorithm (RSA). The output of the proposed tuned recurrent neural network models is compared to the ones of several other state-of-the-art metaheuristic optimization approaches that are applied for the same task, using the same experimental setup, and the obtained results indicate the proposed approach as the better alternative. Moreover, the best recurrent model achieved the best results with R2 of 0.604, and a normalized MSE value of 0.014, which yields an improvement of around 13% over traditional machine learning models.

/pdf/metaheuristic-based-hyperparameter-tuning-for-recurrent-deep-1ftbb203.pdf

Metaheuristic-Based Hyperparameter Tuning for Recurrent Deep Learning: Application to the Prediction of Solar Energy Generation

The number ${U}_{n}$ of equivalence classes of Boolean functions of n variables and the number ${V}_{n}$ of equivalence classes of vectorial Boolean functions of n variables under the action of four groups of transformations are considered. The four groups are the group ${S}_{n}'$ of permutations of variables, the group ${G}_{n}$ of permutations and complementations, the linear group $\mathrm {GL}({n},2)$ and the affine group $\mathrm {AGL}({n},2)$ . Harrison obtained cycle indexes for these groups and the expressions for ${U}_{n}$ and ${V}_{n}$ in terms of the corresponding cycle index. He also tabulated the numbers ${U}_{n}$ , ${V}_{n}$ and the cycle indexes for ${n}\le 6$ for ${S}_{n}'$ and ${G}_{n}$ , and for ${n}\le 5$ for $\mathrm {GL}({n},2)$ and $\mathrm {AGL}({n},2)$ . This bound was only recently slightly exceeded. Fripertinger implemented computation of cycle indexes for $\mathrm {GL}({n},{q})$ and $\mathrm {AGL}({n},{q})$ ; if ${q}=2$ this implementation works for about ${n}\le 21$ . By introducing appropriate precomputed tables, we reduced the cycle index computation to evaluation of a sum over partitions of n for all the four groups. Using this more efficient procedure, we obtained values of ${U}_{n}$ , ${V}_{n}$ and the explicit cycle index expressions for larger values of n .

Miodrag Zivkovic

Papers

Convolutional Neural Networks Hyperparameters Optimization Using Sine Cosine Algorithm

Carrot grading system using computer vision feature parameters and a cascaded graph convolutional neural network

Energy efficient security architecture for wireless sensor networks

Metaheuristic-Based Hyperparameter Tuning for Recurrent Deep Learning: Application to the Prediction of Solar Energy Generation

On the Number of Equivalence Classes of Boolean and Invertible Boolean Functions