Mean Squared Error Loss Function Research Articles

Improving 3D dose prediction for breast radiotherapy using novel glowing masks and gradient-weighted loss functions.

The quality of treatment plans for breast cancer can vary greatly. This variation could be reduced by using dose prediction to automate treatment planning. Our work investigates novel methods for training deep-learning models that are capable of producing high-quality dose predictions for breast cancer treatment planning. The goal of this work was to compare the performance impact of two novel techniques for deep learning dose prediction models for tangent field treatments for breast cancer. The first technique, a "glowing" mask algorithm, encodes the distance from a contour into each voxel in a mask. The second, a gradient-weighted mean squared error (MSE) loss function, emphasizes the error in high-dose gradient regions in the predicted image. Four 3D U-Net deep learning models were trained using the planning CT and contours of the heart, lung, and tumor bed as inputs. The dataset consisted of 305 treatment plans split into 213/46/46 training/validation/test sets using a 70/15/15% split. We compared the impact of novel "glowing" anatomical mask inputs and a novel gradient-weighted MSE loss function to their standard counterparts, binary anatomical masks, and MSE loss, using an ablation study methodology. To assess performance, we examined the mean error and mean absolute error (ME/MAE) in dose across all within-body voxels, the error in mean dose to heart, ipsilateral lung, and tumor bed, dice similarity coefficient (DSC) across isodose volumes defined by 0%-100% prescribed dose thresholds, and gamma analysis (3%/3mm). The combination of novel glowing masks and gradient weighted loss function yielded the best-performing model in this study. This model resulted in a mean ME of 0.40%, MAE of 2.70%, an error in mean dose to heart and lung of -0.10 and 0.01Gy, and an error in mean dose to the tumor bed of -0.01%. The median DSC at 50/95/100% isodose levels were 0.91/0.87/0.82. The mean 3D gamma pass rate (3%/3mm) was 93%. This study found the combination of novel anatomical mask inputs and loss function for dose prediction resulted in superior performance to their standard counterparts. These results have important implications for the field of radiotherapy dose prediction, as the methods used here can be easily incorporated into many other dose prediction models for other treatment sites. Additionally, this dose prediction model for breast radiotherapy has sufficient performance to be used in an automated planning pipeline for tangent field radiotherapy and has the major benefit of not requiring a PTV for accurate dose prediction.

Speech Enhancement Using Joint DNN-NMF Model Learned with Multi-Objective Frequency Differential Spectrum Loss Function

We propose a multi-objective joint model of non-negative matrix factorization (NMF) and deep neural network (DNN) with a new loss function for speech enhancement. The proposed loss function (LMOFD) is a weighted combination of a frequency differential spectrum mean squared error (MSE)-based loss function (LFD) and a multi-objective MSE loss function LMO. The conventional MSE loss function computes the discrepancy between the estimated speech and clean speech across all frequencies, disregarding the process of changing amplitude in the frequency domain which contains valuable information. The differential spectrum representation retains spectral peaks that carry important information. Using this representation helps to ensure that this information in the speech signal is reserved. Also, on the other hand, noise spectra typically have a flat shape and as the differential operation makes the flat spectral partly close to zero, the differential spectrum is resistant to noises with smooth structures. Thus, we propose using a frequency-differentiated loss function that considers the magnitude spectrum differentiations between the neighboring frequency bins in each time frame. This approach maintains the spectrum variations of the objective signal in the frequency domain, which can effectively reduce the noise deterioration effects. The multi-objective MSE term LMO is a combined two-loss function related to the NMF coefficients which are the intermediate output targets, and the original spectral signals as the actual output targets. The use of encoded NMF coefficients as low-dimensional structural features for DNN serves as prior knowledge and helps the learning process. LMO is used beside LFD to take advantage of both the properties of the original and the differential spectrum in the training loss function. Moreover, a DNN-based noise classification and fusion strategy (NCF) is proposed to exploit a discriminative model for noise reduction. The experiments reveal the improvements of the proposed approach compared to the previous methods.

Deep learning models for groundwater level prediction based on delay penalty

Abstract In irrigation agriculture, predicting groundwater level (GWL) using deep learning models can help decision-makers coordinate surface water and groundwater usage, thus aiding in the sustainable development and utilization of groundwater. However, when making a long sequence prediction, prediction sequences often have severe delays affecting the availability of prediction results. In this paper, a new loss function is proposed to minimize the lag and oversmoothing on the prediction of GWLs. GWL, meteorology, and pumping data are collected via an irrigation Internet of Things system in Hutubi County, Xinjiang. Through Pearson's correlation analysis, historical potential evapotranspiration (ET0), groundwater extraction, and GWL were chosen to predict GWLs. Datasets were constructed through the proposed spatiotemporal data fusion method; then, the best model from the six deep learning models was selected by comparing the prediction capability of the datasets. Finally, the mean-squared error (MSE) loss function is replaced by the proposed loss function. Compared to the mean absolute error, MSE, and predicted sequence graphs, the new loss function significantly depresses the time delay with similar prediction accuracy.

Downscaling of surface wind forecasts using convolutional neural networks

Abstract. Near-surface winds over complex terrain generally feature a large variability at the local scale. Forecasting these winds requires high-resolution numerical weather prediction (NWP) models, which drastically increase the duration of simulations and hinder them in running on a routine basis. Nevertheless, downscaling methods can help in forecasting such wind flows at limited numerical cost. In this study, we present a statistical downscaling of WRF (Weather Research and Forecasting) wind forecasts over southeastern France (including the southwestern part of the Alps) from its original 9 km resolution onto a 1 km resolution grid (1 km NWP model outputs are used to fit our statistical models). Downscaling is performed using convolutional neural networks (CNNs), which are the most powerful machine learning tool for processing images or any kind of gridded data, as demonstrated by recent studies dealing with wind forecast downscaling. The previous studies mostly focused on testing new model architectures. In this study, we aimed to extend these works by exploring different output variables and their associated loss function. We found that there is no one approach that outperforms the others in terms of both the direction and the speed at the same time. Finally, the best overall performance is obtained by combining two CNNs, one dedicated to the direction forecast based on the calculation of the normalized wind components using a customized mean squared error (MSE) loss function and the other dedicated to the speed forecast based on the calculation of the wind components and using another customized MSE loss function. Local-scale, topography-related wind features, which were poorly forecast at 9 km, are now well reproduced, both for speed (e.g., acceleration on the ridge, leeward deceleration, sheltering in valleys) and direction (deflection, valley channeling). There is a general improvement in the forecast, especially during the nighttime stable stratification period, which is the most difficult period to forecast. The result is that, after downscaling, the wind speed bias is reduced from −0.55 to −0.01 m s−1, the wind speed MAE is reduced from 1.02 to 0.69 m s−1 (32 % reduction) and the wind direction MAE is reduced from 25.9 to 15.5∘ (40 % reduction) in comparison with the 9 km resolution forecast.

A mathematical model predicting the abundance and spatio-temporal dispersal of adfluvial salmonid (Salvelinus malma) fry in a stream

Models describing the spatial and temporal distributions of fishes are necessary for conservation and management of populations and their habitats. The early life stages of fishes are a particularly important period that can influence population dynamics and fisheries productivity. The fry of riverine migratory fish species are particularly influenced by downstream transport although relatively few models have been used to predict spatio-temporal abundance and dispersion of fry in combination with life history parameters in unperturbed habitats. We present a novel model predicting the downstream dispersal of age-0 adfluvial Dolly Varden in a spawning stream draining into Kronotskoye Lake basin, Kamchatka Peninsula, Russia. The model accounted for the number of and distance between known spawning grounds, diel activity, daily fish growth, and intra-specific diversity (i.e., morphotype) to describe how fry abundance changed at a specific stream location over the course of spring and summer. Data were collected using a trapnet monitored periodically during spring and summer to record changes in abundance, size (growth), and life history composition (‘nosed’ or ‘white’ morphotype) of fish between age-0 and -2. The experimental data were fitted to the model using a gradient descent optimization algorithm and a mean squared error loss function was used to verify how well experimental data fit to the theoretical model. The predicted abundance of age-0 fish was 18,831 with a mean dispersion rate of ∼36 m/day and a transit time between spawning area and stream mouth ranging 32–76 days. The catch was mainly comprised (81%) of the slower growing ‘nosed’ morphotype and exhibited variation in abundance over time and decreased considerably by the end of summer. Our comprehensive model will be useful to simulate the downstream dispersal of salmonid fishes whose early life stages inhabit riverine systems.

A physics-guided eXtreme gradient boosting model for predicting the initial productivity of oil wells

The initial productivity of oil wells is an important index in oilfield development. Due to the nonlinear relationship among the numerous affecting factors of productivity, productivity prediction methods based on seepage theory are difficult to characterize this nonlinear relationship. The machine learning method provides a novel theory to solve this problem. However, the data-driven machine learning method neglects the physical-based seepage theory. For balancing the advantage of the seepage model and the data-driven model, the productivity formula was used to guide the machine learning algorithm for achieving nonlinear regression in this study. Therefore, an innovative physics-guided eXtreme Gradient Boosting (XGBoost) trees algorithm was proposed to predict the initial productivity. The productivity formula was employed as the physical constraint and combined with the mean-square error (MSE) loss function of the XGBoost algorithm. A hyperparameter was introduced to balance the productivity formula and the MSE formula for constructing this adaptive physics-guided loss function. Moreover, the random subsampling cross-validation approach was adopted to train the XGBoost model. The grid search approach was introduced to optimize the hyperparameters of the model. The results show that based on the numerical simulator data, the initial productivity prediction accuracy of physics-guided XGBoost was verified as 94.67%, which was higher than the completely data-driven model. Moreover, the robustness of the novel adaptive physics-guided loss function for improving forecasting accuracy had been validated with sample sizes between 50 and 20000. By introducing the physics-guided sampling process and loss function, the applicability of the XGBoost algorithm has been extended to small sample conditions. Furthermore, the practical accuracy of physics-guided XGBoost was verified as 84.74% by 162 oil wells in the Penglai oilfield. Therefore, the application value of combining the physical knowledge with the XGBoost model has been confirmed in oilfield situations.

TGPred: efficient methods for predicting target genes of a transcription factor by integrating statistics, machine learning and optimization.

Four statistical selection methods for inferring transcription factor (TF)-target gene (TG) pairs were developed by coupling mean squared error (MSE) or Huber loss function, with elastic net (ENET) or least absolute shrinkage and selection operator (Lasso) penalty. Two methods were also developed for inferring pathway gene regulatory networks (GRNs) by combining Huber or MSE loss function with a network (Net)-based penalty. To solve these regressions, we ameliorated an accelerated proximal gradient descent (APGD) algorithm to optimize parameter selection processes, resulting in an equally effective but much faster algorithm than the commonly used convex optimization solver. The synthetic data generated in a general setting was used to test four TF-TG identification methods, ENET-based methods performed better than Lasso-based methods. Synthetic data generated from two network settings was used to test Huber-Net and MSE-Net, which outperformed all other methods. The TF-TG identification methods were also tested with SND1 and gl3 overexpression transcriptomic data, Huber-ENET and MSE-ENET outperformed all other methods when genome-wide predictions were performed. The TF-TG identification methods fill the gap of lacking a method for genome-wide TG prediction of a TF, and potential for validating ChIP/DAP-seq results, while the two Net-based methods are instrumental for predicting pathway GRNs.

Fitting Imbalanced Uncertainties in Multi-output Time Series Forecasting

We focus on multi-step ahead time series forecasting with the multi-output strategy. From the perspective of multi-task learning (MTL), we recognize imbalanced uncertainties between prediction tasks of different future time steps. Unexpectedly, trained by the standard summed Mean Squared Error (MSE) loss, existing multi-output forecasting models may suffer from performance drops due to the inconsistency between the loss function and the imbalance structure. To address this problem, we reformulate each prediction task as a distinct Gaussian Mixture Model (GMM) and derive a multi-level Gaussian mixture loss function to better fit imbalanced uncertainties in multi-output time series forecasting. Instead of using the two-step Expectation-Maximization (EM) algorithm, we apply the self-attention mechanism on the task-specific parameters to learn the correlations between different prediction tasks and generate the weight distribution for each GMM component. In this way, our method jointly optimizes the parameters of the forecasting model and the mixture model simultaneously in an end-to-end fashion, avoiding the need of two-step optimization. Experiments on three real-world datasets demonstrate the effectiveness of our multi-level Gaussian mixture loss compared to models trained with the standard summed MSE loss function. All the experimental data and source code are available at https://github.com/smallGum/GMM-FNN .

Rapid deep-learning-assisted design method for 2-bit coding metasurfaces.

This paper proposes a deep-learning-assisted design method for 2-bit coding metasurfaces. This method uses a skip connection module and the idea of an attention mechanism in squeeze-and-excitation networks based on a fully connected network and a convolutional neural network. The accuracy limit of the basic model is further improved. The convergence ability of the model increased nearly 10 times, and the mean-square error loss function converges to 0.000168. The forward prediction accuracy of the deep-learning-assisted model is 98%, and the accuracy of inverse design results is 97%. This approach offers the advantages of an automatic design process, high efficiency, and low computational cost. It can serve users who lack metasurface design experience.

Loss Function Design for Data-Driven Predictors to Enhance the Energy Efficiency of Connected and Automated Vehicles

In car-following scenarios, accurate previews of the preceding vehicle’s trajectory are essential for minimizing the energy consumption of a following automated vehicle. This work presents a novel design strategy for data-driven vehicle speed predictors to increase the energy efficiency of the following connected and automated vehicle. The loss function is formulated as a weighted-mean-squared error, where the weights are tuned based on the influence of uncertainty at an individual prediction step on the energy consumption of the automated following vehicle. The efficacy of the proposed loss function is validated by applying it to training representative predictors and testing the predictors in the energy-optimal control of the following vehicle. The energy saving of the optimal controller is evaluated by comparing the electricity consumption of a battery electric vehicle with the human driver following, emulated by an intelligent driver model. Simulation results show that the energy consumption of an electric vehicle is reduced by an average of 12% compared to a human driver following demonstrated by an intelligent driver model, whereas using a conventional loss function, a mean-squared error loss function, reduces by 10%.

Tunable Neural Networks for Multi-Material Image Formation from Spectral CT Measurements.

Quantitative estimation of multi-material density images is an important goal for Spectral CT imaging. However, material decomposition is a poorly-conditioned nonlinear inverse problem. Maximum-likelihood model-based material decomposition results in very noisy material density image estimates. One increasingly popular strategy for noise reduction is to apply deep neural networks for multi-material image formation. The most common loss function is mean squared error with respect to supervised target images such as ground truth or higher-dose cases. However, we believe that the mean-squared error loss function has several issues for multi-material image formation. In this work, we present a new loss function which includes multiple noise realizations with separate weights on covariance and bias for joint denoising of all material bases. By modulating these weights, it is possible to tune the image quality of neural network output images. To demonstrate our proposed approach, we conducted a simulation of a water/calcium/gadolinium spectral CT imaging scenario using a deep neural network for multi-material image denoising. Our results show that by changing the weights of our proposed loss function, it is possible to control the tradeoff between variance and bias for individual materials as well as the control over the bias coupling between materials.

Raindrop Size Distribution Prediction by an Improved Long Short-Term Memory Network

The observation of and research on raindrop size distribution (DSD) is important for mastering and understanding the mutual restriction relationship between cloud dynamics and cloud microphysics in a process of precipitation; it also plays an irreplaceable role in many fields, such as radar meteorology, weather modification, boundary layer land surface processes, aerosols, etc. Using more than 1.7 million minutes of raindrop data observed with 17 laser disdrometers at 17 stations in Anhui Province, China, from 7 August 2009 to 30 April 2020, a DSD training dataset was constructed. Furthermore, the data are fitted to a normalized Gamma function and used to obtain its three parameters, i.e., the normalized intercept Nw, the mass weighted average diameter Dm, and the shape factor μ. Based on the long short-term memory network (LSTM), a DSD Gamma distribution prediction network (DSDnet) was designed. In the process of modeling based on DSDnet, a self-defined loss function (SLF) was proposed in order to improve the DSD prediction by increasing the weight values in the poor fitting regions according to the common mean square error loss function (MLF). By means of the training dataset, a DSDnet-based model was trained to realize the prediction of Nw, Dm, and μ minute-to-minute over the course of 30 min, and then was evaluated by the test dataset according to three indicators, namely, mean relative error (MRE), mean absolute error (MAE), and correlation coefficient (CC). The CC of lgNw, Dm, and μ can reach 0.93403, 0.90934, and 0.89741 for 12-min predictions, and 0.87559, 0.85261, and 0.84564 for 30-min predictions, respectively, which means that the DSD prediction accuracy within 30 min can basically reach the application level. Furthermore, the 12- and 30-min predictions of 3 precipitation processes were taken as examples to fully demonstrate the application effect of model. The prediction effects of Nw and Dm are better than that of μ, and the stratiform precipitation is better than the convective and convective-stratiform mixed cloud precipitation.

Adera2.0: A Drug Repurposing Workflow for Neuroimmunological Investigations Using Neural Networks.

Drug repurposing in the context of neuroimmunological (NI) investigations is still in its primary stages. Drug repurposing is an important method that bypasses lengthy drug discovery procedures and focuses on discovering new usages for known medications. Neuroimmunological diseases, such as Alzheimer’s, Parkinson’s, multiple sclerosis, and depression, include various pathologies that result from the interaction between the central nervous system and the immune system. However, the repurposing of NI medications is hindered by the vast amount of information that needs mining. We previously presented Adera1.0, which was capable of text mining PubMed for answering query-based questions. However, Adera1.0 was not able to automatically identify chemical compounds within relevant sentences. To challenge the need for repurposing known medications for neuroimmunological diseases, we built a deep neural network named Adera2.0 to perform drug repurposing. The workflow uses three deep learning networks. The first network is an encoder and its main task is to embed text into matrices. The second network uses a mean squared error (MSE) loss function to predict answers in the form of embedded matrices. The third network, which constitutes the main novelty in our updated workflow, also uses a MSE loss function. Its main usage is to extract compound names from relevant sentences resulting from the previous network. To optimize the network function, we compared eight different designs. We found that a deep neural network consisting of an RNN neural network and a leaky ReLU could achieve 0.0001 loss and 67% sensitivity. Additionally, we validated Adera2.0’s ability to predict NI drug usage against the DRUG Repurposing Hub database. These results establish the ability of Adera2.0 to repurpose drug candidates that can shorten the development of the drug cycle. The workflow could be download online.

Improved Grip Force Prediction Using a Loss Function that Penalizes Reward Related Neural Information.

Neural activity in the sensorimotor cortices has been previously shown to correlate with kinematics, kinetics, and non-sensorimotor variables, such as reward. In this work, we compare the grip force offline Brain Machine Interface (BMI) prediction performance, of a simple artificial neural network (ANN), under two loss functions: the standard mean squared error (MSE) and a modified reward penalized mean squared error (RP_MSE), which penalizes for correlation between reward and grip force. Our results show that the ANN performs significantly better under the RP_MSE loss function in three brain regions: dorsal premotor cortex (PMd), primary motor cortex (M1) and the primary somatosensory cortex (S1) by approximately 6%.

Attention-aware 3D U-Net convolutional neural network for knowledge-based planning 3D dose distribution prediction of head-and-neck cancer.

PurposeDeep learning–based knowledge‐based planning (KBP) methods have been introduced for radiotherapy dose distribution prediction to reduce the planning time and maintain consistent high‐quality plans. This paper presents a novel KBP model using an attention‐gating mechanism and a three‐dimensional (3D) U‐Net for intensity‐modulated radiation therapy (IMRT) 3D dose distribution prediction in head‐and‐neck cancer.MethodsA total of 340 head‐and‐neck cancer plans, representing the OpenKBP—2020 AAPM Grand Challenge data set, were used in this study. All patients were treated with the IMRT technique and a dose prescription of 70 Gy. The data set was randomly divided into 64%/16%/20% as training/validation/testing cohorts. An attention‐gated 3D U‐Net architecture model was developed to predict full 3D dose distribution. The developed model was trained using the mean‐squared error loss function, Adam optimization algorithm, a learning rate of 0.001, 120 epochs, and batch size of 4. In addition, a baseline U‐Net model was also similarly trained for comparison. The model performance was evaluated on the testing data set by comparing the generated dose distributions against the ground‐truth dose distributions using dose statistics and clinical dosimetric indices. Its performance was also compared to the baseline model and the reported results of other deep learning‐based dose prediction models.ResultsThe proposed attention‐gated 3D U‐Net model showed high capability in accurately predicting 3D dose distributions that closely replicated the ground‐truth dose distributions of 68 plans in the test set. The average value of the mean absolute dose error was 2.972 ± 1.220 Gy (vs. 2.920 ± 1.476 Gy for a baseline U‐Net) in the brainstem, 4.243 ± 1.791 Gy (vs. 4.530 ± 2.295 Gy for a baseline U‐Net) in the left parotid, 4.622 ± 1.975 Gy (vs. 4.223 ± 1.816 Gy for a baseline U‐Net) in the right parotid, 3.346 ± 1.198 Gy (vs. 2.958 ± 0.888 Gy for a baseline U‐Net) in the spinal cord, 6.582 ± 3.748 Gy (vs. 5.114 ± 2.098 Gy for a baseline U‐Net) in the esophagus, 4.756 ± 1.560 Gy (vs. 4.992 ± 2.030 Gy for a baseline U‐Net) in the mandible, 4.501 ± 1.784 Gy (vs. 4.925 ± 2.347 Gy for a baseline U‐Net) in the larynx, 2.494 ± 0.953 Gy (vs. 2.648 ± 1.247 Gy for a baseline U‐Net) in the PTV_70, and 2.432 ± 2.272 Gy (vs. 2.811 ± 2.896 Gy for a baseline U‐Net) in the body contour. The average difference in predicting the D 99 value for the targets (PTV_70, PTV_63, and PTV_56) was 2.50 ± 1.77 Gy. For the organs at risk, the average difference in predicting the Dmax (brainstem, spinal cord, and mandible) and Dmean (left parotid, right parotid, esophagus, and larynx) values was 1.43 ± 1.01 and 2.44 ± 1.73 Gy, respectively. The average value of the homogeneity index was 7.99 ± 1.45 for the predicted plans versus 5.74 ± 2.95 for the ground‐truth plans, whereas the average value of the conformity index was 0.63 ± 0.17 for the predicted plans versus 0.89 ± 0.19 for the ground‐truth plans. The proposed model needs less than 5 s to predict a full 3D dose distribution of 64 × 64 × 64 voxels for a new patient that is sufficient for real‐time applications.ConclusionsThe attention‐gated 3D U‐Net model demonstrated a capability in predicting accurate 3D dose distributions for head‐and‐neck IMRT plans with consistent quality. The prediction performance of the proposed model was overall superior to a baseline standard U‐Net model, and it was also competitive to the performance of the best state‐of‐the‐art dose prediction method reported in the literature. The proposed model could be used to obtain dose distributions for decision‐making before planning, quality assurance of planning, and guiding‐automated planning for improved plan consistency, quality, and planning efficiency.

Cost-oriented load forecasting

Accurate load prediction is an effective way to reduce power system operation costs. Traditionally, the Mean Square Error (MSE) is a common-used loss function to guide the training of an accurate load forecasting model. However, the MSE loss function is unable to precisely reflect the real costs associated with forecasting errors because the cost caused by forecasting errors in the real power system is probably neither symmetric nor quadratic. To tackle this issue, this paper proposes a generalized cost-oriented load forecasting framework. Specifically, how to obtain a differentiable loss function that reflects real cost and how to integrate the loss function with regression models are studied. The economy and effectiveness of the proposed load forecasting method are verified by the case studies of an optimal dispatch problem that is built on the IEEE 30-bus system and the open load dataset from the Global Energy Forecasting Competition 2012(GEFCom2012).

A novel loss function of deep learning in wind speed forecasting

Wind speed forecasting is an essential task in improving the efficiency of the energy supply. Currently, deep learning models have become extremely popular, where the traditional mean squared error (MSE) loss function is often employed. Unfortunately, the MSE loss function cannot accurately measure the nonlinearity of wind speed data, and new loss functions have seldom been developed for wind speed forecasting. In addition, the MSE loss function is sensitive to outliers, degrading the stability. To address these problems, we propose a kernel MSE loss function to evaluate the ubiquitous nonlinearity of deep learning errors in the reproducing kernel Hilbert space. The new loss function utilizes the kernel skills in the loss function of deep learning methods for the first time. The first and second derivatives of the new loss function guarantee the robustness against outliers. The experimental results with three fundamental deep learning methods on three public datasets validate that the new loss function is efficient and superior in most cases, and its resultant error can be 95% smaller than MSE in multiple step prediction. The results imply that developing a loss function with kernel skills is a new way to get better results.

Deep Neural Network Based Electrical Impedance Tomographic Sensing Methodology for Large-Area Robotic Tactile Sensing

Electrical impedance tomography (EIT) based tactile sensor offers significant benefits on practical deployment because of its sparse electrode allocation, including durability, large-area scalability, and low fabrication cost, but the degradation of a tactile spatial resolution has remained challenging. This article describes a deep neural network based EIT reconstruction framework, the EIT neural network (EIT-NN), alleviating this tradeoff between tactile sensing performance and hardware simplicity. EIT-NN learns a computationally efficient, nonlinear reconstruction attribute, achieving high-resolution tactile sensation and well-generalized reconstruction capability to address arbitrary complex touch modalities. We train EIT-NN by presenting a sim-to-real dataset synthesis strategy for computationally efficient generalizability. Furthermore, we propose a spatial sensitivity aware mean-squared error loss function, which uses an intrinsic spatial sensitivity of the sensor to guarantee a well-posed EIT operation. We validate an outperformance of EIT-NN against conventional EIT sensing methods by conducting a simulation study, a single-touch indentation test, and a two-point discrimination test. The results show improved spatial resolution, sensitivity, and localization accuracy. The beneficial features of the generalized sensing of EIT-NN were demonstrated by examining touch modality discrimination performance.

Image Inpainting Using Double Discriminator Generative Adversarial Networks

The image inpainting algorithms based on GAN have problems such as color inconsistencies of the inpainted region with the surrounding part and the mode collapse during the training. This paper proposes an inpainting algorithm based on double discriminator generative adversarial networks (GANs). This algorithm introduces the loss function of WGAN-GP into a global discriminator and local discriminator independently. It combines the Mean-Square Error (MSE) and feature reconstruction loss function to train the inpainting model and repair the missing area. The algorithm uses more dilated convolutions instead of standard convolutions to obtain a larger receptive field and the skip connection to enhance the structural prediction ability of the generator. The experiment results on the labeled faces in the wild dataset show that our algorithm improves the quality of image inpainting and the stability of the training process.

Predicting the Risk of Chronic Kidney Disease (CKD) Using Machine Learning Algorithm

Background: Creatinine is a type of metabolite of blood that is strongly correlated to glomerular filtration rate (GFR). As measuring GFR is difficult, creatinine value is used for indirectly determining GFR and then the stage of chronic kidney disease (CKD). Adding a creatinine test into routine health examination could detect CKD. As more items for comprehensive examination means higher cost, creatinine testing is not included in the routine health examination in many countries. An algorithm based on common test results, without creatinine test, to evaluate the risk of CKD will increase the chance of its early detection and treatment. Methods: In this study, we used open source data containing 1 million samples. These data contain 23 health-related features, including common diagnostic test results provided by National Health Insurance Sharing Service (NHISS). A low GFR indicates possible chronic kidney disease (CKD). As is commonly accepted in the medical community, a GFR of 60 mL/min is used as the threshold, below which is considered to have CKD. In this study, the first step aims to build a regression model to predict the value of creatinine from 23 features, and then combine the predicted value of creatinine with the original 23 features to evaluate the risk of CKD. We will show by simulation that by the proposed method we can achieve better prediction results compared to direct prediction from 23 features. The data is extremely unbalanced for predicting the target variable creatinine. We used undersampling method and proposed a new cost-sensitive mean-squared error (MSE) loss function to deal with the problem. Regrading model selection, this work used three machine learning models: a bagging tree model named Random Forest, a boosting tree model named XGBoost, and a neural network based model named ResNet. To improve the result of the creatinine predictor, we averaged results from eight predictors, a method known as ensemble learning. Finally, the predicted creatinine and the original 23 features is used to predict the risk of CKD. Results: We optimized results of R-Squared (R2) value to select the appropriate undersampling strategy and the regression model for the regression stage of creatinine prediction. Ensembled model achieved the best performance of R2 of 0.5590. The six factors from 23 are selected from the top of the list of how strongly they affect the creatinine value. They are sex, age, hemoglobin, the level of urine protein, waist circumference, and habit of smoking. Using the predicted value of creatinine, an area under Receiver Operating Characteristic curve (AUC) of 0.76 is achieved while classifying samples for CKD. Conclusions: Using commonly available health parameters, the proposed system can assess the risk of CKD for public health. High-risk subjects can be screened and advised to take a creatinine test for further confirmation. In this way, we can reduce the impact of CKD on public health and facilitate early detection for many, where a blanket test of creatinine is not available for all.