Day-ahead energy production in small hydropower plants: uncertainty-aware forecasts through effective coupling of knowledge and data

Motivated by the challenges induced by the so-called Target Model and the associated changes to the current structure of the energy market, we revisit the problem of day-ahead prediction of power production from Small Hydropower Plants (SHPPs) without storage capacity. Using as an example a typical run-of-river SHPP in Western Greece, we test alternative forecasting schemes (from regression-based to machine learning) that take advantage of different levels of information. In this respect, we investigate whether it is preferable to use as predictor the known energy production of previous days, or to predict the day-ahead inflows and next estimate the resulting energy production via simulation. Our analyses indicate that the second approach becomes clearly more advantageous when the expert's knowledge about the hydrological regime and the technical characteristics of the SHPP is incorporated within the model training procedure. Beyond these, we also focus on the predictive uncertainty that characterize such forecasts, with overarching objective to move beyond the standard, yet risky, point forecasting methods, providing a single expected value of power production. Finally, we discuss the use of the proposed forecasting procedure under uncertainty in the real-world electricity market.

Assignment of Focus Position with Convolutional Neural Networks in Adaptive Lens Based Axial Scanning for Confocal Microscopy

Adaptive lenses offer axial scanning without mechanical translation and thus are promising to replace mechanical-movement-based axial scanning in microscopy. The scan is accomplished by sweeping the applied voltage. However, the relation between the applied voltage and the resulting axial focus position is not unambiguous. Adaptive lenses suffer from hysteresis effects, and their behaviour depends on environmental conditions. This is especially a hurdle when complex adaptive lenses are used that offer additional functionalities and are controlled with more degrees of freedom. In such case, a common approach is to iterate the voltage and monitor the adaptive lens. Here, we introduce an alternative approach which provides a single shot estimation of the current axial focus position by a convolutional neural network. We use the experimental data of our custom confocal microscope for training and validation. This leads to fast scanning without photo bleaching of the sample and opens the door to automatized and aberration-free smart microscopy. Applications in different types of laser-scanning microscopes are possible. However, maybe the training procedure of the neural network must be adapted for some use cases.

Deep metric attention learning for skin lesion classification in dermoscopy images

Currently, convolutional neural networks (CNNs) have made remarkable achievements in skin lesion classification because of their end-to-end feature representation abilities. However, precise skin lesion classification is still challenging because of the following three issues: (1) insufficient training samples, (2) inter-class similarities and intra-class variations, and (3) lack of the ability to focus on discriminative skin lesion parts. To address these issues, we propose a deep metric attention learning CNN (DeMAL-CNN) for skin lesion classification. In DeMAL-CNN, a triplet-based network (TPN) is first designed based on deep metric learning, which consists of three weight-shared embedding extraction networks. TPN adopts a triplet of samples as input and uses the triplet loss to optimize the embeddings, which can not only increase the number of training samples, but also learn the embeddings robust to inter-class similarities and intra-class variations. In addition, a mixed attention mechanism considering both the spatial-wise and channel-wise attention information is designed and integrated into the construction of each embedding extraction network, which can further strengthen the skin lesion localization ability of DeMAL-CNN. After extracting the embeddings, three weight-shared classification layers are used to generate the final predictions. In the training procedure, we combine the triplet loss with the classification loss as a hybrid loss to train DeMAL-CNN. We compare DeMAL-CNN with the baseline method, attention methods, advanced challenge methods, and state-of-the-art skin lesion classification methods on the ISIC 2016 and ISIC 2017 datasets, and test its generalization ability on the PH2 dataset. The results demonstrate its effectiveness.

Management Game

Digital disruption and continuous productivity improvement require more from people management, thus raising the bar for leadership competencies. International studies indicate that leadership competence gaps are large and traditional leadership training methods does not seem to solve this problem. This article's findings supports this situation. The authors will open the complexity behind organizational productivity development and present game theoretical architecture that simulates management behavior effects to human performance. New methods enable practice-based learning that enables formatting leaders' behavior so that it will create long-term success with continuous change. The authors will present gamified leadership training procedure and discuss the practical learning experiences from a management simulation game. The authors' study reveals challenges at interactive leadership skills, thus, it is argued, that there seems to be problems at the leadership mind-set. Therefore, more sophisticated learning methods and tools should be used.

IMPROVING ENGLISH SKILLS OF YOUTH ORGANIZATIONS IN PAMMANA VILLAGE THROUGH AUDIO-VISUAL MEDIA

Pammana is a village in Kabupaten Wajo, Sulawesi Selatan, where the majority of the residents work as farmers. With a low income, parents' capacity to send their children to a higher level of education is always hampered. Based on the aforementioned circumstances, the writer and team are interested in delivering English language training for the youth organization utilizing an audio-visual medium. The goal of this program was to make the training sessions more exciting and accessible to the participants through the use of audiovisual media. This English training exercise has a positive impact on the Pammana village youth group. These favorable effects might be seen as a result of the training. The learning results of the training participants improved after only a few months of following the training procedure using the audio-visual medium. Before attending the training, the trainees thought English was a difficult language to learn; however, after attending the training utilizing audio-visual media, the trainees began to feel more at ease, and they thought learning English through audio-visual media was enjoyable.

Development of object state evaluation method in intelligent decision support systems

Accurate and objective object analysis requires multi-parameter estimation with significant computational costs. A methodological approach to improve the accuracy of assessing the state of the monitored object is proposed. This methodological approach is based on a combination of fuzzy cognitive models, advanced genetic algorithm and evolving artificial neural networks. The methodological approach has the following sequence of actions: building a fuzzy cognitive model; correcting the fuzzy cognitive model and training knowledge bases. The distinctive features of the methodological approach are that the type of data uncertainty and noise is taken into account while constructing the state of the monitored object using fuzzy cognitive models. The novelties while correcting fuzzy cognitive models using a genetic algorithm are taking into account the type of data uncertainty, taking into account the adaptability of individuals to iteration, duration of the existence of individuals and topology of the fuzzy cognitive model. The advanced genetic algorithm increases the efficiency of correcting factors and the relationships between them in the fuzzy cognitive model. This is achieved by finding solutions in different directions by several individuals in the population. The training procedure consists in learning the synaptic weights of the artificial neural network, the type and parameters of the membership function and the architecture of individual elements and the architecture of the artificial neural network as a whole. The use of the method allows increasing the efficiency of data processing at the level of 16–24 % using additional advanced procedures. The proposed methodological approach should be used to solve the problems of assessing complex and dynamic processes characterized by a high degree of complexity.

Inspecting Buildings Using Drones and Computer Vision: A Machine Learning Approach to Detect Cracks and Damages

Manual inspection of infrastructure damages such as building cracks is difficult due to the objectivity and reliability of assessment and high demands of time and costs. This can be automated using unmanned aerial vehicles (UAVs) for aerial imagery of damages. Numerous computer vision-based approaches have been applied to address the limitations of crack detection but they have their limitations that can be overcome by using various hybrid approaches based on artificial intelligence (AI) and machine learning (ML) techniques. The convolutional neural networks (CNNs), an application of the deep learning (DL) method, display remarkable potential for automatically detecting image features such as damages and are less sensitive to image noise. A modified deep hierarchical CNN architecture has been used in this study for crack detection and damage assessment in civil infrastructures. The proposed architecture is based on 16 convolution layers and a cycle generative adversarial network (CycleGAN). For this study, the crack images were collected using UAVs and open-source images of mid to high rise buildings (five stories and above) constructed during 2000 in Sydney, Australia. Conventionally, a CNN network only utilizes the last layer of convolution. However, our proposed network is based on the utility of multiple layers. Another important component of the proposed CNN architecture is the application of guided filtering (GF) and conditional random fields (CRFs) to refine the predicted outputs to get reliable results. Benchmarking data (600 images) of Sydney-based buildings damages was used to test the proposed architecture. The proposed deep hierarchical CNN architecture produced superior performance when evaluated using five methods: GF method, Baseline (BN) method, Deep-Crack BN, Deep-Crack GF, and SegNet. Overall, the GF method outperformed all other methods as indicated by the global accuracy (0.990), class average accuracy (0.939), mean intersection of the union overall classes (IoU) (0.879), precision (0.838), recall (0.879), and F-score (0.8581) values. Overall, the proposed CNN architecture provides the advantages of reduced noise, highly integrated supervision of features, adequate learning, and aggregation of both multi-scale and multilevel features during the training procedure along with the refinement of the overall output predictions.

An Evolutionary Computation Approach to Resource Allocation in Container-based Clouds

<p>A container-based cloud is a new trend in cloud computing that introduces more granular management of cloud resources. Compared with VM-based clouds, container-based clouds can further improve energy efficiency with a finer granularity resource allocation in data centers. The current allocation approaches for VM-based clouds cannot be used in container-based clouds. The first reason is that existing research lacks appropriate models that can represent the interaction of allocation features. Many critical features, such as VM overhead, are also not considered in the current models. The second reason is that current allocation approaches do not perform well to the three frequently encountered allocation scenarios, offline allocation, on-line allocation, and multi-objective allocation. Current approaches for these scenarios are mostly based on greedy heuristics that can be easily stuck at local optimum, or meta-heuristics that consider a simplified one-level allocation problem. Evolutionary Computation (EC) is particularly good at solving combinatorial optimization problems for both off-line and on-line scenarios. The overall goal of this thesis is to propose an EC approach to the three allocation scenarios in order to improve the performance of container-based clouds. Specifically, we aim to optimize energy consumption in all the scenarios. An additional objective, availability of the application, is considered for the multi-objective scenario. First, this thesis investigates two promising representations, vector-based and group-based. We propose two novel vector-based (e.g., Single- chromosome Genetic Algorithm (SGA) and Dual-chromosome GA (DGA)) and a group-based GA approaches for the off-line allocation scenario. Cor- responding genetic operators and decoding processes are also developed and evaluated. Two contributions have been made. Firstly, a novel offline model has been proposed based on current models with additional features. It can be used to evaluate allocation algorithms. Secondly, two types of problem representation, vector-based and group-based, are investigated and three novel approaches are proposed. The three approaches are compared with state-of-the-art approaches. The results show that all solutions produced by these approaches are better than the state-of-the-art approaches and group-based GA is the best approach. Second, this thesis proposes a novel genetic programming hyper-heuristic (GPHH) and a cooperative coevolution (CCGP)-based approach for the on-line allocation scenario. These hyper-heuristic methods can automatically generate allocation rules. For the GPHH-based approach, we develop a novel training procedure to generate reservation-based rules for allocating containers to VM instances. For the CCGP-based approach, we introduce a new terminal set and develop a training procedure to generate allocation rules for two-level allocations. We analyze both human-designed rules and generated rules to provide insights for algorithm designers. Two contributions have been proposed for the on-line problem. First, the on-line model for the on-line allocation scenario is developed. Second, a novel terminal set and training procedures are developed. The automatically generated heuristics perform significantly better than the manually designed heuristics. Third, this thesis proposes a multi-objective approach that generates a set of trade-off solutions for the cloud providers to choose from. Our novel approach, namely Nondominated Sorting-Group GA (NS-GGA), combines the group-based representation and the NSGA-II framework. The experimental results are compared with existing approaches. The results show that our proposed NS-GGA approach outperforms all other approaches. We propose two novelties. The first novelty is the multi-objective model including objectives of energy consumption and availability. The second novelty is the NS-GGA approach that combines the group-based representation with NSGA-II. The allocation solutions found by NS-GGA dominate the solutions found by other existing approaches.</p>

An Evolutionary Computation Approach to Resource Allocation in Container-based Clouds

<p>A container-based cloud is a new trend in cloud computing that introduces more granular management of cloud resources. Compared with VM-based clouds, container-based clouds can further improve energy efficiency with a finer granularity resource allocation in data centers. The current allocation approaches for VM-based clouds cannot be used in container-based clouds. The first reason is that existing research lacks appropriate models that can represent the interaction of allocation features. Many critical features, such as VM overhead, are also not considered in the current models. The second reason is that current allocation approaches do not perform well to the three frequently encountered allocation scenarios, offline allocation, on-line allocation, and multi-objective allocation. Current approaches for these scenarios are mostly based on greedy heuristics that can be easily stuck at local optimum, or meta-heuristics that consider a simplified one-level allocation problem. Evolutionary Computation (EC) is particularly good at solving combinatorial optimization problems for both off-line and on-line scenarios. The overall goal of this thesis is to propose an EC approach to the three allocation scenarios in order to improve the performance of container-based clouds. Specifically, we aim to optimize energy consumption in all the scenarios. An additional objective, availability of the application, is considered for the multi-objective scenario. First, this thesis investigates two promising representations, vector-based and group-based. We propose two novel vector-based (e.g., Single- chromosome Genetic Algorithm (SGA) and Dual-chromosome GA (DGA)) and a group-based GA approaches for the off-line allocation scenario. Cor- responding genetic operators and decoding processes are also developed and evaluated. Two contributions have been made. Firstly, a novel offline model has been proposed based on current models with additional features. It can be used to evaluate allocation algorithms. Secondly, two types of problem representation, vector-based and group-based, are investigated and three novel approaches are proposed. The three approaches are compared with state-of-the-art approaches. The results show that all solutions produced by these approaches are better than the state-of-the-art approaches and group-based GA is the best approach. Second, this thesis proposes a novel genetic programming hyper-heuristic (GPHH) and a cooperative coevolution (CCGP)-based approach for the on-line allocation scenario. These hyper-heuristic methods can automatically generate allocation rules. For the GPHH-based approach, we develop a novel training procedure to generate reservation-based rules for allocating containers to VM instances. For the CCGP-based approach, we introduce a new terminal set and develop a training procedure to generate allocation rules for two-level allocations. We analyze both human-designed rules and generated rules to provide insights for algorithm designers. Two contributions have been proposed for the on-line problem. First, the on-line model for the on-line allocation scenario is developed. Second, a novel terminal set and training procedures are developed. The automatically generated heuristics perform significantly better than the manually designed heuristics. Third, this thesis proposes a multi-objective approach that generates a set of trade-off solutions for the cloud providers to choose from. Our novel approach, namely Nondominated Sorting-Group GA (NS-GGA), combines the group-based representation and the NSGA-II framework. The experimental results are compared with existing approaches. The results show that our proposed NS-GGA approach outperforms all other approaches. We propose two novelties. The first novelty is the multi-objective model including objectives of energy consumption and availability. The second novelty is the NS-GGA approach that combines the group-based representation with NSGA-II. The allocation solutions found by NS-GGA dominate the solutions found by other existing approaches.</p>

Noise-trained deep neural networks effectively predict human vision and its neural responses to challenging images

Deep neural networks (DNNs) for object classification have been argued to provide the most promising model of the visual system, accompanied by claims that they have attained or even surpassed human-level performance. Here, we evaluated whether DNNs provide a viable model of human vision when tested with challenging noisy images of objects, sometimes presented at the very limits of visibility. We show that popular state-of-the-art DNNs perform in a qualitatively different manner than humans—they are unusually susceptible to spatially uncorrelated white noise and less impaired by spatially correlated noise. We implemented a noise training procedure to determine whether noise-trained DNNs exhibit more robust responses that better match human behavioral and neural performance. We found that noise-trained DNNs provide a better qualitative match to human performance; moreover, they reliably predict human recognition thresholds on an image-by-image basis. Functional neuroimaging revealed that noise-trained DNNs provide a better correspondence to the pattern-specific neural representations found in both early visual areas and high-level object areas. A layer-specific analysis of the DNNs indicated that noise training led to broad-ranging modifications throughout the network, with greater benefits of noise robustness accruing in progressively higher layers. Our findings demonstrate that noise-trained DNNs provide a viable model to account for human behavioral and neural responses to objects in challenging noisy viewing conditions. Further, they suggest that robustness to noise may be acquired through a process of visual learning.