Artificial Neural Networks and Deep Learning for Genomic Prediction of Continuous Outcomes

This chapter provides elements for implementing deep neural networks (deep learning) for continuous outcomes. We give details of the hyperparameters to be tuned in deep neural networks and provide a general guide for doing this task with more probability of success. Then we explain the most popular deep learning frameworks that can be used to implement these models as well as the most popular optimizers available in many software programs for deep learning. Several practical examples with plant breeding data for implementing deep neural networks in the Keras library are outlined. These examples take into account many components in the predictor as well many hyperparameters (hidden layer, number of neurons, learning rate, optimizers, penalization, etc.) for which we also illustrate how the tuning process can be done to increase the probability of a successful application.

Hyperparameter Tuning for Machine Learning Algorithms Used for Arabic Sentiment Analysis

Machine learning models are used today to solve problems within a broad span of disciplines. If the proper hyperparameter tuning of a machine learning classifier is performed, significantly higher accuracy can be obtained. In this paper, a comprehensive comparative analysis of various hyperparameter tuning techniques is performed; these are Grid Search, Random Search, Bayesian Optimization, Particle Swarm Optimization (PSO), and Genetic Algorithm (GA). They are used to optimize the accuracy of six machine learning algorithms, namely, Logistic Regression (LR), Ridge Classifier (RC), Support Vector Machine Classifier (SVC), Decision Tree (DT), Random Forest (RF), and Naive Bayes (NB) classifiers. To test the performance of each hyperparameter tuning technique, the machine learning models are used to solve an Arabic sentiment classification problem. Sentiment analysis is the process of detecting whether a text carries a positive, negative, or neutral sentiment. However, extracting such sentiment from a complex derivational morphology language such as Arabic has been always very challenging. The performance of all classifiers is tested using our constructed dataset both before and after the hyperparameter tuning process. A detailed analysis is described, along with the strengths and limitations of each hyperparameter tuning technique. The results show that the highest accuracy was given by SVC both before and after the hyperparameter tuning process, with a score of 95.6208 obtained when using Bayesian Optimization.

Coupling Matrix Extraction of Microwave Filters by Using One-Dimensional Convolutional Autoencoders

The tuning of microwave filter is important and complex. Extracting coupling matrix from given S-parameters is a core task for filter tuning. In this article, one-dimensional convolutional autoencoders (1D-CAEs) are proposed to extract coupling matrix from S-parameters of narrow-band cavity filter and apply this method to the computer-aided tuning process. The training of 1D-CAE model consists of two steps. First, in the encoding part, one-dimensional convolutional neural network (1D-CNN) with several convolution layers and pooling layers is used to extract the coupling matrix from the S-parameters during the microwave filters’ tuning procedure. Second, in the decoding part, several full connection layers are employed to reconstruct the S-parameters to ensure the accuracy of extraction. The S-parameters obtained by measurement or simulation exist with phase shift, so the influence of phase shift must be removed. The efficiency of the presented method in this article is validated by a sixth-order cross-coupled filter simulation model tuning example.

Plug-and-Play Few-shot Object Detection with Meta Strategy and Explicit Localization Inference

Aiming at recognizing and localizing the object of novel categories by a few reference samples, few-shot object detection is a quite challenging task. Previous works often depend on the fine-tuning process to transfer their model to the novel category and rarely consider the defect of fine-tuning, resulting in many drawbacks. For example, these methods are far from satisfying in the low-shot or episode-based scenarios since the fine-tuning process in object detection requires much time and high-shot support data. To this end, this paper proposes a plug-and-play few-shot object detection (PnP-FSOD) framework that can accurately and directly detect the objects of novel categories without the fine-tuning process. To accomplish the objective, the PnP-FSOD framework contains two parallel techniques to address the core challenges in the few-shot learning, i.e., across-category task and few-annotation support. Concretely, we first propose two simple but effective meta strategies for the box classifier and RPN module to enable the across-category object detection without fine-tuning. Then, we introduce two explicit inferences into the localization process to reduce its dependence on the annotated data, including explicit localization score and semi-explicit box regression. In addition to the PnP-FSOD framework, we propose a novel one-step tuning method that can avoid the defects in fine-tuning. It is noteworthy that the proposed techniques and tuning method are based on the general object detector without other prior methods, so they are easily compatible with the existing FSOD methods. Extensive experiments show that the PnP-FSOD framework has achieved the state-of-the-art few-shot object detection performance without any tuning method. After applying the one-step tuning method, it further shows a significant lead in both efficiency, precision, and recall, under varied few-shot evaluation protocols.

Plug-and-Play Few-shot Object Detection with Meta Strategy and Explicit Localization Inference

Aiming at recognizing and localizing the object of novel categories by a few reference samples, few-shot object detection is a quite challenging task. Previous works often depend on the fine-tuning process to transfer their model to the novel category and rarely consider the defect of fine-tuning, resulting in many drawbacks. For example, these methods are far from satisfying in the low-shot or episode-based scenarios since the fine-tuning process in object detection requires much time and high-shot support data. To this end, this paper proposes a plug-and-play few-shot object detection (PnP-FSOD) framework that can accurately and directly detect the objects of novel categories without the fine-tuning process. To accomplish the objective, the PnP-FSOD framework contains two parallel techniques to address the core challenges in the few-shot learning, i.e., across-category task and few-annotation support. Concretely, we first propose two simple but effective meta strategies for the box classifier and RPN module to enable the across-category object detection without fine-tuning. Then, we introduce two explicit inferences into the localization process to reduce its dependence on the annotated data, including explicit localization score and semi-explicit box regression. In addition to the PnP-FSOD framework, we propose a novel one-step tuning method that can avoid the defects in fine-tuning. It is noteworthy that the proposed techniques and tuning method are based on the general object detector without other prior methods, so they are easily compatible with the existing FSOD methods. Extensive experiments show that the PnP-FSOD framework has achieved the state-of-the-art few-shot object detection performance without any tuning method. After applying the one-step tuning method, it further shows a significant lead in both efficiency, precision, and recall, under varied few-shot evaluation protocols.

Untilted edge slot antennas excited by T-shaped wires

An effective method is proposed to excite the untilted edge slot antennas. In the proposed method, two T-shaped wires are placed at both sides of each untilted slot. Two legs of each T-shaped wire are connected to the waveguide walls, while the third leg is open. One of the T-shaped wires connects the upper broad wall of the waveguide to the narrow wall of the waveguide which contains the slots. Similarly, the other T-shaped wires connect the lower broad wall of the waveguide to the narrow wall of the waveguide that contains the slots. The phase reversal between the adjacent slots can be accomplished by changing the orientation of the T-shaped wires. It is shown that the arm lengths of the connected wires can be employed as a parameter to control the radiated power by the slots. Furthermore, the arm lengths of the open wires can be selected properly to control the dynamic range of the equivalent normalized susceptance associated with each untilted slot antenna. To validate the effectiveness of the proposed antennas, two linear arrays consisting of 11 slots have been designed, implemented, and tested. The simulation and the measurement results of the designed arrays show that for the proposed untilted edge slot antennas, the dynamic range of the equivalent normalized susceptance is improved significantly which leads to a straightforward design process with no requirement to resort to any time-consuming tuning process.

Demonstration of Thermally Tunable Multi-Band and Ultra-Broadband Metamaterial Absorbers Maintaining High Efficiency during Tuning Process

Metamaterial absorbers (MMAs) with dynamic tuning features have attracted great attention recently, but most realizations to date have suffered from a decay in absorptivity as the working frequency shifts. Here, thermally tunable multi-band and ultra-broadband MMAs based on vanadium dioxide (VO2) are proposed, with nearly no reduction in absorption during the tuning process. Simulations demonstrated that the proposed design can be switched between two independently designable multi-band frequency ranges, with the absorptivity being maintained above 99.8%. Moreover, via designing multiple adjacent absorption spectra, an ultra-broadband switchable MMA that maintains high absorptivity during the tuning process is also demonstrated. Raising the ambient temperature from 298 K to 358 K, the broadband absorptive range shifts from 1.194–2.325 THz to 0.398–1.356 THz, while the absorptivity remains above 90%. This method has potential for THz communication, smart filtering, detecting, imaging, and so forth.

RHOASo: An Early Stop Hyper-Parameter Optimization Algorithm

This work proposes a new algorithm for optimizing hyper-parameters of a machine learning algorithm, RHOASo, based on conditional optimization of concave asymptotic functions. A comparative analysis of the algorithm is presented, giving particular emphasis to two important properties: the capability of the algorithm to work efficiently with a small part of a dataset and to finish the tuning process automatically, that is, without making explicit, by the user, the number of iterations that the algorithm must perform. Statistical analyses over 16 public benchmark datasets comparing the performance of seven hyper-parameter optimization algorithms with RHOASo were carried out. The efficiency of RHOASo presents the positive statistically significant differences concerning the other hyper-parameter optimization algorithms considered in the experiments. Furthermore, it is shown that, on average, the algorithm needs around 70% of the iterations needed by other algorithms to achieve competitive performance. The results show that the algorithm presents significant stability regarding the size of the used dataset partition.

A PID Tuning Strategy Based on a Variable Weight Beetle Antennae Search Algorithm for Hydraulic Systems

To select reasonable PID controller parameters and improve control performances of hydraulic systems, a variable weight beetle antenna search algorithm is proposed for PID tuning in the hydraulic system. The beetle antennae search algorithm is inspired by the beetle preying habit depending on symmetry antennae on the head. The proposed algorithm added the exponential equation mechanism strategy in the basic algorithm to further improve the searching performance, the convergence speed, and the optimization accuracy and obtain new iteration and an updating method in the global searching and local searching stages. In the PID tuning process, advantages of less parameters and fast iteration are realized in the PID tuning process. In this paper, different dimension functions were tested, and results calculated by the proposed algorithm were compared with other famous algorithms, and the numerical analysis was carried out, including the iteration, the box-plot, and the searching path, which comprehensively showed the searching balance in the proposed algorithm. Finally, the reasonable PID controller parameters are found by using the proposed method, and the tuned PID controller is introduced into the hydraulic system for control, and the time-domain response characteristics and frequency response characteristics are given. The results show that the proposed PID tuning method has good PID parameter tuning ability, and the tuned PID has a good control ability, which makes the hydraulic system achieve the desired effect.