Active Exploration by Chance-Constrained Optimization for Voltage Regulation with Reinforcement Learning

Voltage regulation in distribution networks encounters a challenge of handling uncertainties caused by the high penetration of photovoltaics (PV). This research proposes an active exploration (AE) method based on reinforcement learning (RL) to respond to the uncertainties by regulating the voltage of a distribution network with battery energy storage systems (BESS). The proposed method integrates engineering knowledge to accelerate the training process of RL. The engineering knowledge is the chance-constrained optimization. We formulate the problem in a chance-constrained optimization with a linear load flow approximation. The optimization results are used to guide the action selection of the exploration for improving training efficiency and reducing the conserveness characteristic. The comparison of methods focuses on how BESSs are used, training efficiency, and robustness under varying uncertainties and BESS sizes. We implement the proposed algorithm, a chance-constrained optimization, and a traditional Q-learning in the IEEE 13 Node Test Feeder. Our evaluation shows that the proposed AE method has a better response to the training efficiency compared to traditional Q-learning. Meanwhile, the proposed method has advantages in BESS usage in conserveness compared to the chance-constrained optimization.

Performance analysis of the TensorFlow library with different optimisation algorithms

This paper presents the results of performance analysis of the Tensorflow library used in machine learning and deep neural networks. The analysis focuses on comparing the parameters obtained when training the neural network model for optimization algorithms: Adam, Nadam, AdaMax, AdaDelta, AdaGrad. Special attention has been paid to the differences between the training efficiency on tasks using microprocessor and graphics card. For the study, neural network models were created in order to recognise Polish handwritten characters. The results obtained showed that the most efficient algorithm is AdaMax, while the computer component used during the research only affects the training time of the neural network model used.

The Supervised Learning Model for Analyzing the Sportsperson Training Efficiency

The driving concept of students’ sports training involves a unique activity that is often tightly correlated to students’ efficiency and varies with the momentum of sports training. Supervised learning is one of the smart methods with positive results in the fields of classification techniques. Due to the excessive currency unit associated with sports, sports forecasting is a growing area that must be well predicted. Therefore, in this paper, sports training based on the supervised learning (STSLM) model has been proposed to evaluate and predict student sports efficiency. STSLM models are based on various variables, such as traditional student ratings, performance, and efficiency. The emphasis is on the efficiency of students predicting sports outcomes. STSLM defines evaluation methods, information sources, effective models for testing students’ sports training, and unique challenges to forecast sports outcomes. The experimental results have been performed. The suggested STSLM model enhances the efficiency ratio of 96.3%, injury prevention level of 98.2%, fitness level of 95.5%, evaluation ratio of 98.8%, and training optimization ratio of 97.2% compared to other existing approaches.

Efficiency Evaluation Model of Students' Outdoor Taichi Training Based on Supervised Learning

Taichi has a long history and spread widely in China. It has played an important role in maintaining human health. This paper studies the efficiency evaluation model of students' outdoor Taichi training based on supervised learning. This paper expounds the key points and importance of sports health preservation through the collation of sports health preservation related literature, the comparison between traditional sports methods and modern sports methods, and the interpretation of knowledge and practice Taichi. This paper analyzes the influence of Taichi on the total score of body symptoms. Based on supervised learning algorithm, this paper compares Taichi with modern sports and irregular sports, studies the effect of Taichi on body conditioning, and puts forward an evaluation model of Taichi training efficiency. The results obtained in this paper lay a theoretical foundation for the promotion of knowledge and practice Taichi, and provide data reference for the establishment of Taichi training efficiency.

A Novel Technique for Image Steganalysis Based on Separable Convolution and Adversarial Mechanism

Image steganalysis is a technique for detecting the presence of hidden information in images, which has profound significance for maintaining cyberspace security. In recent years, various deep steganalysis networks have been proposed in academia, and have achieved good detection performance. Although convolutional neural networks (CNNs) can effectively extract the features describing the image content, the difficulty lies in extracting the subtle features that describe the existence of hidden information. Considering this concern, this paper introduces separable convolution and adversarial mechanism, and proposes a new network structure that effectively solves the problem. The separable convolution maximizes the residual information by utilizing its channel correlation. The adversarial mechanism makes the generator extract more content features to mislead the discriminator, thus separating more steganographic features. We conducted experiments on BOSSBase1.01 and BOWS2 to detect various adaptive steganography algorithms. The experimental results demonstrate that our method extracts the steganographic features effectively. The separable convolution increases the signal-to-noise ratio, maximizes the channel correlation of residuals, and improves efficiency. The adversarial mechanism can separate more steganographic features, effectively improving the performance. Compared with the traditional steganalysis methods based on deep learning, our method shows obvious improvements in both detection performance and training efficiency.

Design and Realization of Shooting Virtual Training System for UAV-borne Missile

UAV-borne missile is effective weapon to attack enemy ground targets. It is expensive, costly and difficult to live-fire drill. Using virtual training instead of actual training can greatly improve the training efficiency and the combat effectiveness. The article regards the operation training of a certain type of UAV-borne missile shooting training as the research object, based on the development of a visual simulation system for UAV-borne missile, uses the object-oriented design method to design a virtual training system based on LabVIEW. The system can realize the shooting operation training of trainees in a virtual environment, and achieve the goals of reduce training costs; improve training efficiency and shorten training period.

Task shifting of intravitreal injections from physicians to nurses: a qualitative study

Background Intravitreal injections of anti-vascular endothelial growth factor are high-volume procedures and represent a considerable workload on ophthalmology departments. Several departments have tried to meet this increase by shifting the task to nurses. To maintain high-quality patient care, we developed a training program for nurses that certifies them to administer injections. This qualitative study aimed to evaluate whether the nurses were confident and in control after participating in the training program and whether they were satisfied with the training and the new task. Methods Between 2014 and 2018, 12 registered nurses were trained in a tertiary hospital in central Norway. All the nurses were interviewed, either individually (n = 7) or in a group (n = 5). We analysed the interviews using Graneheim and Lundman’s qualitative content analysis. Results Eight subthemes were clustered within four main themes: 1) procedure and challenges, 2) motivation, 3) cooperation and confidence, and 4) evaluation. The nurses felt confident and in control when administering injections but experienced moments of insecurity. The new task gave the nurses a sense of achievement, and they highlighted improvement of patients’ lives as positive. A greater level of responsibility gave the nurses pride in their profession. They had suggestions that could improve training efficiency but were overall satisfied with the training program. Conclusions Our study showed that the nurses were satisfied with the training and that learning a new task led to higher self-esteem and increased respect from patients and colleagues. Suggestions to improve the training were identified; these should be considered before implementation by other departments.

Automated Robust Interpretation of Intraoperative Electrophysiological Signals – A Bayesian Deep Learning Approach

Intraoperative neurophysiological monitoring (IONM) is an essential tool during numerous surgical interventions to assess and monitor the functional integrity of neural structures at risk. A reliable signal interpretation is of importance to support medical staff by reducing manual evaluation. Deep learning (DL) techniques proved to be a robust tool for the analysis of neurophysiological data. The large amount of required manually labeled data as well as the lack of interpretability of the results however often limit the use of DL in medical scenarios. A possible way to tackle these obstacles is the utilization of Bayesian deep learning (BDL) methods. The modelling of uncertainties in the network parameters and the thereby possible quantification of predictive uncertainties allows both the identification of potential erroneous predictions as well as the targeted selection of informative signals in the context of active learning. To evaluate the applicability of BDL for the analysis of electrophysiological data as well as to increase the training efficiency by active learning, we implemented a multi-task Bayesian Convolutional Neural Network (BCNN) for the simultaneous classification of action potentials and the assessment of relevant signal characteristics (latency, maximum, minimum). We compare the results for electromyographical signals (EMG), containing in total approximately twelve thousand signals from 34 patients, with both a traditional non-Bayesian single-task and multi-task CNN. For all models, including the BCNN, we could achieve similar performances with detection rates over 97% accuracy. Further, we could improve training efficiency of the BCNN using pool-based active learning and therefore significantly reduce the required amount of manual labeling. The evaluated predictive uncertainties of the BCNN prove useful both for the efficient selection of informative signals in the context of active learning as well as the interpretation of the predictive posterior distribution and therefore trustworthiness of the classifications.