In the last few years, the deep learning (DL) computing paradigm has been deemed the Gold Standard in the machine learning (ML) community. Moreover, it has gradually become the most widely used computational approach in the field of ML, thus achieving outstanding results on several complex cognitive tasks, matching or even beating those provided by human performance. One of the benefits of DL is the ability to learn massive amounts of data. The DL field has grown fast in the last few years and it has been extensively used to successfully address a wide range of traditional applications. More importantly, DL has outperformed well-known ML techniques in many domains, e.g., cybersecurity, natural language processing, bioinformatics, robotics and control, and medical information processing, among many others. Despite it has been contributed several works reviewing the State-of-the-Art on DL, all of them only tackled one aspect of the DL, which leads to an overall lack of knowledge about it. Therefore, in this contribution, we propose using a more holistic approach in order to provide a more suitable starting point from which to develop a full understanding of DL. Specifically, this review attempts to provide a more comprehensive survey of the most important aspects of DL and including those enhancements recently added to the field. In particular, this paper outlines the importance of DL, presents the types of DL techniques and networks. It then presents convolutional neural networks (CNNs) which the most utilized DL network type and describes the development of CNNs architectures together with their main features, e.g., starting with the AlexNet network and closing with the High-Resolution network (HR.Net). Finally, we further present the challenges and suggested solutions to help researchers understand the existing research gaps. It is followed by a list of the major DL applications. Computational tools including FPGA, GPU, and CPU are summarized along with a description of their influence on DL. The paper ends with the evolution matrix, benchmark datasets, and summary and conclusion.

Deep learning requires a large amount of data to perform well. However, the field of medical image analysis suffers from a lack of sufficient data for training deep learning models. Moreover, medical images require manual labeling, usually provided by human annotators coming from various backgrounds. More importantly, the annotation process is time-consuming, expensive, and prone to errors. Transfer learning was introduced to reduce the need for the annotation process by transferring the deep learning models with knowledge from a previous task and then by fine-tuning them on a relatively small dataset of the current task. Most of the methods of medical image classification employ transfer learning from pretrained models, e.g., ImageNet, which has been proven to be ineffective. This is due to the mismatch in learned features between the natural image, e.g., ImageNet, and medical images. Additionally, it results in the utilization of deeply elaborated models. In this paper, we propose a novel transfer learning approach to overcome the previous drawbacks by means of training the deep learning model on large unlabeled medical image datasets and by next transferring the knowledge to train the deep learning model on the small amount of labeled medical images. Additionally, we propose a new deep convolutional neural network (DCNN) model that combines recent advancements in the field. We conducted several experiments on two challenging medical imaging scenarios dealing with skin and breast cancer classification tasks. According to the reported results, it has been empirically proven that the proposed approach can significantly improve the performance of both classification scenarios. In terms of skin cancer, the proposed model achieved an F1-score value of 89.09% when trained from scratch and 98.53% with the proposed approach. Secondly, it achieved an accuracy value of 85.29% and 97.51%, respectively, when trained from scratch and using the proposed approach in the case of the breast cancer scenario. Finally, we concluded that our method can possibly be applied to many medical imaging problems in which a substantial amount of unlabeled image data is available and the labeled image data is limited. Moreover, it can be utilized to improve the performance of medical imaging tasks in the same domain. To do so, we used the pretrained skin cancer model to train on feet skin to classify them into two classes—either normal or abnormal (diabetic foot ulcer (DFU)). It achieved an F1-score value of 86.0% when trained from scratch, 96.25% using transfer learning, and 99.25% using double-transfer learning.

The main aim of this paper is to solve a path planning problem for an autonomous mobile robot in static and dynamic environments. The problem is solved by determining the collision-free path that satisfies the chosen criteria for shortest distance and path smoothness. The proposed path planning algorithm mimics the real world by adding the actual size of the mobile robot to that of the obstacles and formulating the problem as a moving point in the free-space. The proposed algorithm consists of three modules. The first module forms an optimized path by conducting a hybridized Particle Swarm Optimization-Modified Frequency Bat (PSO-MFB) algorithm that minimizes distance and follows path smoothness criteria. The second module detects any infeasible points generated by the proposed PSO-MFB Algorithm by a novel Local Search (LS) algorithm integrated with the PSO-MFB algorithm to be converted into feasible solutions. The third module features obstacle detection and avoidance (ODA), which is triggered when the mobile robot detects obstacles within its sensing region, allowing it to avoid collision with obstacles. The simulation results indicate that this method generates an optimal feasible path even in complex dynamic environments and thus overcomes the shortcomings of conventional approaches such as grid methods. Moreover, compared to recent path planning techniques, simulation results show that the proposed PSO-MFB algorithm is highly competitive in terms of path optimality.

Planning an optimal path for a mobile robot is a complicated problem as it allows the mobile robots to navigate autonomously by following the safest and shortest path between starting and goal points. The present work deals with the design of intelligent path planning algorithms for a mobile robot in static and dynamic environments based on swarm intelligence optimization. A modification based on the age of the ant is introduced to standard ant colony optimization, called aging-based ant colony optimization (ABACO). The ABACO was implemented in association with grid-based modeling for the static and dynamic environments to solve the path planning problem. The simulations are run in the MATLAB environment to test the validity of the proposed algorithms. Simulations showed that the proposed path planning algorithms result in superior performance by finding the shortest and the most free-collision path under various static and dynamic scenarios. Furthermore, the superiority of the proposed algorithms was proved through comparisons with other traditional path planning algorithms with different static environments.

The design of a swarm optimization-based fractional control for engineering application is an active research topic in the optimization analysis. This work offers the analysis, design, and simulation of a new neural network- (NN) based nonlinear fractional control structure. With suitable arrangements of the hidden layer neurons using nonlinear and linear activation functions in the hidden and output layers, respectively, and with appropriate connection weights between different hidden layer neurons, a new class of nonlinear neural fractional-order proportional integral derivative (NNFOPID) controller is proposed and designed. It is obtained by approximating the fractional derivative and integral actions of the FOPID controller and applied to the motion control of nonholonomic differential drive mobile robot (DDMR). The proposed NNFOPID controller’s parameters consist of derivative, integral, and proportional gains in addition to fractional integral and fractional derivative orders. The tuning of these parameters makes the design of such a controller much more difficult than the classical PID one. To tackle this problem, a new swarm optimization algorithm, namely, MAPSO-EFFO algorithm, has been proposed by hybridization of the modified adaptive particle swarm optimization (MAPSO) and the enhanced fruit fly optimization (EFFO) to tune the parameters of the NNFOPID controller. Firstly, we developed a modified adaptive particle swarm optimization (MAPSO) algorithm by adding an initial run phase with a massive number of particles. Secondly, the conventional fruit fly optimization (FFO) algorithm has been modified by increasing the randomness in the initialization values of the algorithm to cover wider searching space and then implementing a variable searching radius during the update phase by starting with a large radius which decreases gradually during the searching phase. The tuning of the parameters of the proposed NNFOPID controller is carried out by reducing the MS error of 0.000059, whereas the MSE of the nonlinear neural system (NNPID) is equivalent to 0.00079. The NNFOPID controller also decreased control signals that drive DDMR motors by approximately 45 percent compared to NNPID and thus reduced energy consumption in circular trajectories. The numerical simulations revealed the excellent performance of the designed NNFOPID controller by comparing its performance with that of nonlinear neural (NNPID) controllers on the trajectory tracking of the DDMR with different trajectories as study cases.