Search

Species interactions play a fundamental role in ecosystems. However, few ecological communities have complete data describing such interactions, which is an obstacle to understanding how ecosystems function and respond to perturbations. Because it is often impractical to collect empirical data for all interactions in a community, various methods have been developed to infer interactions. Machine learning is increasingly being used for making interaction predictions, with random forest being one of the most frequently used of these methods. However, performance of random forest in inferring predator‐prey interactions in terrestrial vertebrates and its sensitivity to training data quality remain untested. We examined predator–prey interactions in two diverse, primarily terrestrial vertebrate classes: birds and mammals. Combining data from a global interaction dataset and a specific community (Simpson Desert, Australia), we tested how well random forest predicted predator–prey interactions for mammals and birds using species' ecomorphological and phylogenetic traits. We also tested how variation in training data quality – manipulated by removing records and switching interaction records to non‐interactions – affected model performance. We found that random forest could predict predator–prey interactions for birds and mammals using ecomorphological or phylogenetic traits, correctly predicting up to 88 and 67% of interactions and non‐interactions in the global and community‐specific datasets, respectively. These predictions were accurate even when there were no records in the training data for focal species. In contrast, false non‐interactions for focal predators in training data strongly degraded model performance. Our results demonstrate that random forest can identify predator–prey interactions for birds and mammals that have few or no interaction records. Furthermore, our study provides guidance on how to prepare training data to optimise machine learning classifiers for predicting species interactions, which could help ecologists 1) address knowledge gaps and explore network‐related questions in data‐poor situations, and 2) predict interactions for range‐expanding species. [ABSTRACT FROM AUTHOR]

Parametrization in bridge construction for the application of artificial intelligence For the application of digital methods in the execution phase, the acquisition of exisiting structures or the construction progress with the aid of LiDAR is a suitable method to subsequently detect components as automatically as possible and e. g., to evaluate them for construction progress monitoring. Machine learning methods are suitable for component detection and are used for various classification problems. One problem is the semantic segmentation of point clouds, in which the individual points are assigned to specific component classes. To solve the segmentation with sufficient accuracy and to make predictions on real data, a large amount of training data is needed. This process is also called "Scan to BIM". In this paper, synthetic data for training a convolutionalneural network (CNN) is provided in a highly parameterized and fully automatic way. An approach for generating variations of bridge models based with the visual programming interface Dynamo in Autodesk Revit is proposed. At the beginning, guidelines for the definition of parameters are evaluated to subsequently use them in adaptive templates. Attributes are assigned to each component and grouped into a feature structure when exported to the IFC data format. This approach resulted in numerous variations of bridge models that can be used for semantic segmentation of point clouds and training of a CNN. [ABSTRACT FROM AUTHOR]

In our recent review paper aiming to introduce deep learning to ecologists, we presented a workflow describing the steps required to create a deep learning model. This figure did not present some of the following steps of model use such as model verification.By ensuring model adequacy, model verification is an important step after model creation in order to answer ecological questions.Adding model verification to a deep learning model development workflow can raise some new issues such as detecting the difference among the multiple datasets or what to do when model verification fails. In the spirit of our previous review, we identify some questions users trying to verify their deep learning model can have and try to find, for each, a solution to help them navigate the steps of deep learning model testing.We provide an additional cheat sheet to quickly help answer common questions regarding using model verification and deep learning. We hope these resources help stimulate further synthesis and coherence in the use of deep learning models in ecology. [ABSTRACT FROM AUTHOR]

Underwater image enhancement algorithms improve image quality and indirectly enhance underwater visibility. Although many underwater image enhancement neural networks have been proposed, they require large amounts of data. To reduce the amount of data required while providing better image enhancement, this study proposes an underwater image colour transfer generative adversarial network (UCT‐GAN). The authors first design a non‐linear mapping function to generate colour cast images according to original images. Then, the authors utilise these image pairs (i.e. colour cast images and corresponding original images) to guide the UCT‐GAN in learning the inverse function of the designed non‐linear mapping function. Finally, colour cast images are restored via the inverse function. A data augmentation method based on Poisson fusion and block combination is also proposed to overcome the problem of requiring a large amount of training data. Moreover, the authors extend UCT‐GAN into a multi‐class colour transfer network to achieve an array of underwater image enhancements. Experimental results indicate that the proposed UCT‐GAN can more effectively resolve underwater image colour cast compared to existing algorithms. [ABSTRACT FROM AUTHOR]

In recent years, image fusion methods based on deep networks have been proposed to combine infrared and visible images for achieving better fusion image. However, issues such as limited training data, scarce reference images and misalignment of multi‐source images, still limit the fusion performance. To address these problems, we propose an end‐to‐end shallow convolutional neural network with structural constraints, which has only one convolutional layer to fuse infrared and visible images. Different from other methods, our proposed model requires less training data and reference images and is more robust to the misalignment of a couple of images. More specifically, the infrared image and the visible image are first provided as inputs to a convolutional layer to extract the information that should be fused; then, all feature maps are concatenated together and fed into a convolutional layer with one channel to obtain the fused image; finally, a structural similarity loss between the fused image and the input infrared and visible images is computed to update the network parameters and eliminate the effects of pixel misalignment. Extensive experiments show the effectiveness of our proposed method on fusion of infrared and visible images with the performance that outperforms the state‐of‐the‐art methods. [ABSTRACT FROM AUTHOR]

For wireless networks, it is highly desirable to design a quality‐of‐service‐promotion strategy to allocate the transmission power. In this study, the authors apply the deep learning methodology to solve such a problem in a duplex device‐to‐device network. The model is formulated as a constrained non‐linear programming problem. The training data are generated by a standard optimisation tool. Then, an artificial neural network is designed to learn the system from the training data. Simulation results show the effectiveness of deep leaning. [ABSTRACT FROM AUTHOR]

Crowding on Mass Rapid Transit (MRT) Systems has been the subject of intense scrutiny from a multitude of academic fields. Crowding prediction can assist system managers in finding effective ways to ease traffic pressure. Forecasting crowding levels is a challenging task due to a series of complex factors. Deep learning models are a common approach often applied to traffic prediction, but due to the imbalance within available data, these models that are based on historical data cannot forecast crowding in Rapid Transit Systems effectively. To solve this problem, this study proposes a model named the weighted resample bidirectional recurrent neural network (WRBRNN). First, the training data were split into different sub‐datasets according to certain predetermined labels. During this training time, every mini‐batch size sequence was weighed and resampled from different sub‐datasets. In this study, the authors carefully arranged the traffic data attributes into several time series, allowing the bidirectional time series information and the model to make reliable predictions. This work performed a case study of the Bay Area Rapid Transit, US system with a year's worth of historical data from 2017. Their results reveal that the model WRBRNN performed well when predicting crowding in an MRT system. [ABSTRACT FROM AUTHOR]

Intervention surgery strongly requires information on the guide‐wire position under the monitoring of X‐ray video. Hence, the related researches such as guide‐wire detecting or tracking have become widespread. However, most of the existing methods require a lot of resources for computing or large data for training since the X‐ray videos have internal physicals such as anatomical skeleton contours and organs that are quite similar to a guide‐wire. This work presents a practical method that only requires a moderate number of training data for detecting a guide‐wire tip in an X‐ray video sequence during the percutaneous coronary intervention surgery. The method applies maximally stable extremal regions (MSER) combine with modified multi‐filters (region area range filter and stroke width variation filter) for region detection and local binary patterns (LBP) for guide‐wire recognition. The motivation for applying MSER and LBP are the robust efficacy and the low requirement of resources. The approach evaluated 20 different sequences of X‐ray videos, a total of 1295 frames. 50 selected frames were used as training templates and others to experiment. The method was successfully performed to the detecting guide‐wires with p‐value < 0.01 compared with conventional MSER methods, 93.7% average detection accuracy, and 21 fps average speed. [ABSTRACT FROM AUTHOR]

Space-time adaptive processing (STAP) for airborne radar with co-prime arrays is shown to have excellent superiority compared to traditional STAP with uniform linear array radar. However, high arithmetic computational complexity and large amount of training data are required in this approach. This motivates the authors to present a new approach which is relatively low computational load and fast convergence with satisfactory performance. Specifically, a reduced-dimension transformation in Doppler domain is incorporated into the existing approach. The reduced-dimension interference covariance matrix and target steering vector are then achieved by performing the reduced-dimension process, and hence two reduced-dimension STAP filters are designed using the derived interference covariance matrix and target-steering vector. Numerical simulations are carried out to reveal the superiority of the proposed approach. [ABSTRACT FROM AUTHOR]

Target detection at sub‐pixel abundances is, in fact, one of the challenging issues of hyperspectral image processing. Selection of optimal bands to improve sub‐pixel target detection (STD) performance is one of the common solutions, applied by many researchers. Nevertheless, the absence of sufficient training data is the main weakness of selecting optimal bands with regard to this approach. The present research introduces a new band selection method for STD in hyperspectral images, based on creating training data, in which the desired target spectrum is implanted randomly in a series of host pixels from the entire hyperspectral image. Afterwards, via running an optimisation algorithm twice, with the aim of minimising the false alarm rate (FAR) in local adaptive coherence estimator target detection algorithm, the number of optimal bands and optimal spectral bands are selected. In this study, the performance of three optimisation methods including the genetic algorithm (GA), Grey Wolf optimisation (GWO), and particle swarm optimisation (PSO) are compared. Experimental results on HyMap and Hyperion datasets show that the proposed method obtains the minimum FAR compared with the rest of the evaluated methods. Also, based on the results obtained, GWO outperforms GA and PSO optimisation methods in the STD domain. [ABSTRACT FROM AUTHOR]

Generative models such as variational autoencoders (VAEs) and generative adversarial networks (GANs) have been recently applied to various fields. However, the VAE and GAN models have blur and mode collapse problems, respectively. Here, the authors propose a novel generative model, self-converging generative network (SCGN), to address the issues. Self-converging means the convergence of latent vectors into themselves through being trained in pairs with training data, by which the SCGN can reconstruct all training data. In the authors' model, the latent vectors and weights of the generator are alternately trained. Specifically, the latent vectors are trained to follow a normal distribution, using a loss function derived from the Kullback–Leibler divergence and a pixel-wise loss. The weights of the generator are adjusted for the generator to produce training data by means of a pixel-wise loss. As a result, their SCGN did not fall into the mode collapse, which occurs in GANs, and made clearer images than VAEs thanks to no use of sampling. Moreover, the SCGN successfully learned the manifold of the dataset in the extensive experiments with CelebA. [ABSTRACT FROM AUTHOR]

Supercapacitors are increasingly applied to the field of electric vehicles. Although the state of charge (SOC) directly shows the remaining capacity of supercapacitors in an energy management system, vehicle drivers also require the changes of supercapacitors in the future for driving reference. How to predict SOC of supercapacitors has become a pressing problem. In order to solve the above problem, a method combining backpropagation (BP) neural network with the Kalman filtering algorithm is proposed to predict SOC in the future. The BP neural network inputs SOC estimated by the Kalman filtering algorithm as the training data to train network, and thereby being able to forecast the SOC in the future period. The algorithm is verified in simulations and experiments under the two conditions: NewYorkBus and NYCC, with the consideration of the influence of the length of training data and temperature. The results show the max absolute error during prediction at different lengths is under 6% in simulations and experiments. In addition, temperature has almost no effect on the prediction accuracy. The current research implies that this method can be applied to predict supercapacitors' SOC in the future. [ABSTRACT FROM AUTHOR]

Manufacturers reduce stray losses in transformers by using magnetic shunts. These losses may cause hot spots in metallic structural parts of transformers such as the tank wall and frames. The calculation of the reduction of the stray losses on the tank wall using magnetic shunts is a complex task. In this study, a method for determining the position of a magnetic shunt and reducing the stray losses on the tank wall of a transformer is presented. Feed‐forward neural networks (FFNNs) and the finite element method are employed. The Latin hypercube sampling is used to achieve a reduced number of training cases. Afterwards, several two‐dimensional finite‐element axisymmetric models of a transformer are solved and used as training data for two FFNN models. A comparison with finite‐element models is carried out to evaluate the proposed method. Finally, a case study is solved using this method. The approach offers an alternative method to predict and reduce the intensity of the stray losses on the tank wall of the transformer in a straightforward way. [ABSTRACT FROM AUTHOR]

This study presents an intelligent solution for load‐frequency control in a restructured power system using a modified traditional frequency response model, suitable for a bilateral‐based deregulation policy. The new approach is based on an extended classifier system with continuous‐valued inputs (XCSR) which is the most successful learning classifier systems. The proposed intelligent solution does not require an accurate model of the system and is more flexible in specifying the control objectives. Also it is an automated learning‐based approach. It means there is not any need to training data and expert knowledge of the system to determine the states and actions, which is a very time‐consuming and difficult stage of designing reinforcement learning‐based solutions. To demonstrate the effectiveness of the proposed method, its performance on a three‐area restructured power system with possible contract scenarios, large load demands and area disturbances has been compared with multi‐agent reinforcement learning‐based controller. The results show that the proposed intelligent solution achieves good robust performance for a wide range of load changes in the presence of system nonlinearities and has good ability to track the contracted and non‐contracted demands. [ABSTRACT FROM AUTHOR]

It is well known that statistical classification procedures should be assessed using data that are separate from those used to train the classifier. This principle is commonly overlooked when the classification procedure in question is population assignment using a set of genetic markers that were chosen specifically on the basis of their allele frequencies from amongst a larger number of candidate markers. This oversight leads to a systematic upward bias in the predicted accuracy of the chosen set of markers for population assignment. Three widely used software programs for selecting markers informative for population assignment suffer from this bias. The extent of this bias is documented through a small set of simulations. The relative effect of the bias is largest when screening many candidate loci from poorly differentiated populations. Simple unbiased methods are presented and their use encouraged. [ABSTRACT FROM AUTHOR]

We consider a two-component mixture model where one component distribution is known while the mixing proportion and the other component distribution are unknown. These kinds of models were first introduced in biology to study the differences in expression between genes. The various estimation methods proposed till now have all assumed that the unknown distribution belongs to a parametric family. In this paper, we show how this assumption can be relaxed. First, we note that generally the above model is not identifiable, but we show that under moment and symmetry conditions some ‘almost everywhere’ identifiability results can be obtained. Where such identifiability conditions are fulfilled we propose an estimation method for the unknown parameters which is shown to be strongly consistent under mild conditions. We discuss applications of our method to microarray data analysis and to the training data problem. We compare our method to the parametric approach using simulated data and, finally, we apply our method to real data from microarray experiments. [ABSTRACT FROM AUTHOR]

Over-learning is a drawback of the error-back-propagation (BP) method for a multilayer feed-forward neural network. The authors have already discussed this problem for the case in which pure training data are available. It was shown that over-learning is caused by the use of a memorization criterion as a substitute for some true criterion that determines generalization ability. A relation between a true criterion and a substitute criterion was used to define mathematically the concept of over-learning, and the concepts of four kinds of admissibility of the substitute criterion in place of the true criterion were introduced. In this paper, we show that the forementioned general framework can also be applied to the case in which training data are noisy. First, the memorization criterion is extended to cover rote-memorization criterion, which requires the same responses as those given by the training data even if they include noise. Next, the framework is applied to the case in which the rote-memorization criterion is used as a substitute for the Wiener criterion. Necessary and sufficient conditions for the four kinds of admissibility are obtained. These conditions lead us to methods for choosing a training set that prevents over-learning. [ABSTRACT FROM AUTHOR]

Gamma tuning is a technique widely used in photography, video image and computer graphics. OLED, also known as organic laser display and organic light‐emitting semiconductor, has developed rapidly in recent years. However, the algorithms related to OLED gamma tuning have developed slowly. At present, OLED gamma tuning algorithm is mainly traditional algorithm, which is difficult to control the adjustment step, which will inevitably increase the number of tuning, resulting in TT difficult to meet the requirements. In order to solve the problem that adjusting step size is difficult to control, a gamma tuning algorithm based on machine learning is proposed in this paper. Getting Initial Value by Find Table Sending[1] and predicted by machine learning algorithm. Experiments show that, compared with the traditional gamma tuning algorithm, the proposed algorithm can effectively shorten the tuning times and thus shorten TT while guaranteeing the tuning effect. [ABSTRACT FROM AUTHOR]