Weighted Loss Function Research Articles

Efficient and effective data imputation with influence functions

Data imputation has been extensively explored to solve the missing data problem. The dramatically rising volume of missing data makes the training of imputation models computationally infeasible in real-life scenarios. In this paper, we propose an efficient and effective data imputation system with influence functions , named EDIT, which quickly trains a parametric imputation model with representative samples under imputation accuracy guarantees. EDIT mainly consists of two modules, i.e., an imputation influence evaluation (IIE) module and a representative sample selection (RSS) module. IIE leverages the influence functions to estimate the effect of (in)complete samples on the prediction result of parametric imputation models. RSS builds a minimum set of the high-effect samples to satisfy a user-specified imputation accuracy. Moreover, we introduce a weighted loss function that drives the parametric imputation model to pay more attention on the high-effect samples. Extensive experiments upon ten state-of-the-art imputation methods demonstrate that, EDIT adopts only about 5% samples to speed up the model training by 4x in average with more than 11% accuracy gain.

Reply to Landry et al. Findings from Diet Comparison Difficult to Interpret in the Absence of Adherence Assessment. Comment on "Tricò et al. Effects of Low-Carbohydrate versus Mediterranean Diets on Weight Loss, Glucose Metabolism, Insulin Kinetics and β-Cell Function in Morbidly Obese Individuals. Nutrients 2021, 13, 1345".

We thank Dr. Landry and colleagues [...].

DW-UNet: Loss Balance under Local-Patch for 3D Infection Segmentation from COVID-19 CT Images.

(1) Background: COVID-19 has been global epidemic. This work aims to extract 3D infection from COVID-19 CT images; (2) Methods: Firstly, COVID-19 CT images are processed with lung region extraction and data enhancement. In this strategy, gradient changes of voxels in different directions respond to geometric characteristics. Due to the complexity of tubular tissues in lung region, they are clustered to the lung parenchyma center based on their filtered possibility. Thus, infection is improved after data enhancement. Then, deep weighted UNet is established to refining 3D infection texture, and weighted loss function is introduced. It changes cost calculation of different samples, causing target samples to dominate convergence direction. Finally, the trained network effectively extracts 3D infection from CT images by adjusting driving strategy of different samples. (3) Results: Using Accuracy, Precision, Recall and Coincidence rate, 20 subjects from a private dataset and eight subjects from Kaggle Competition COVID-19 CT dataset tested this method in hold-out validation framework. This work achieved good performance both in the private dataset (99.94–00.02%, 60.42–11.25%, 70.79–09.35% and 63.15–08.35%) and public dataset (99.73–00.12%, 77.02–06.06%, 41.23–08.61% and 52.50–08.18%). We also applied some extra indicators to test data augmentation and different models. The statistical tests have verified the significant difference of different models. (4) Conclusions: This study provides a COVID-19 infection segmentation technology, which provides an important prerequisite for the quantitative analysis of COVID-19 CT images.

Bayesian Estimation for Two Parameters of Weibull Distribution under Generalized Weighted Loss Function

In this paper, Bayes estimators for the shape and scale parameters of Weibull distribution have been obtained using the generalized weighted loss function, based on Exponential priors. Lindley’s approximation has been used effectively in Bayesian estimation. Based on theMonte Carlo simulation method, those estimators are compared depending on the mean squared errors (MSE’s).

Automatic Pixel-Level Crack Detection for Civil Infrastructure Using Unet++ and Deep Transfer Learning

In the practical applications of crack detection, traditional image processing based detection methods are faced with a lot of challenges, such as low level of intelligence, poor adaptability and so on. In this paper, an automatic pixel-level crack detection method based on deep transfer learning is proposed for engineering applications. The proposed detection method is composed of two stages: crack recognition and crack semantic segmentation, which makes it easy to meet the needs of efficient and reliable detection for large-scale collected images. The fine-tuned Vgg16 model in the first stage can accurately identify the crack images and avoid the computing cost for non-crack images in the further processing. The Unet++ model in the second stage provides a pixel-level semantic segmentation for crack images. Multifold knowledge based transfer learning and weighted loss function are proposed to improve the performance and generalization ability of the models. On 5 publicly available datasets, the proposed method achieved a mIoU of 84.62%. Tested on totally 8064 images with a resolution of 224×224, the proposed method achieved a mean detection speed of 8.1ms per image. Experiment results exhibit that the proposed method can effectively detect the pixel-level cracks. It provides an end-to-end deep learning method for large-scale crack detection for civil infrastructure. For the crack detection in complex scenes and the improvement of processing speed will be our next in-depth study.

Farrerol Induces Cancer Cell Death via ERK Activation in SKOV3 Cells and Attenuates TNF-α-Mediated Lipolysis.

Farrerol (FA) is a flavanone isolated from the Chinese herbal medicine “Man-shan-hong” (Rhododendron dauricum L.). In the present study, FA decreased the viability of SKOV3 cells in a dose- and time-dependent manner, and it induced G2/M cell cycle arrest and cell apoptosis. Cell cycle distribution analysis via flow cytometry showed that FA decreased G1 populations and increased G2/M populations in SKOV3 cells. Additionally, Western blotting confirmed an increase in the expression level of proteins involved in the cell cycle, e.g., CDK and cyclins. FA-induced apoptosis in SKOV3 cells was also investigated using a TUNEL assay, and increased expression levels of proapoptotic factors, including Caspase-3 and poly ADP ribose polymerase (PARP), through the Extracellular signal-regulated kinase (ERK)/MAPK pathway were investigated. Proinflammatory cytokines (e.g., IL-6, TNF-α, and IL-1) have been identified as a driver of the pathological mechanisms underlying involuntary weight loss and impaired physical function, i.e., cachexia, during cancer; in the present study, we showed that farrerol attenuates TNF-α-induced lipolysis and increases adipogenic differentiation in 3T3-L1 cells. Thus, farrerol could potentially be used as an anticancer agent or anticachetic drug.

Bariatric Surgery Improves Renal Function in Patients With Obesity.

BackgroundObesity affects 93.3 million adults in the United States and is a predisposing factor for the development and progression of chronic kidney disease (CKD). The objective of this study is to examine the association between weight loss and renal function in participants undergoing bariatric surgery following a 12-week multidisciplinary, community-based weight loss program.MethodologyThis is a retrospective chart review of participants who voluntarily enrolled in a 12-week multidisciplinary weight loss program prior to bariatric surgery from 2009 to 2018. The primary outcome was to assess the association between weight loss and renal function in participants undergoing bariatric surgery. Secondary outcomes included changes in hemoglobin A1c, lipids, fasting glucose, and blood pressure.ResultsAmong the 55 participants, baseline glomerular filtration rate (GFR) was 49 mL/min/m2, 80% were female, and the average baseline weight was 131 kg. At one-year post-intervention, 69% of patients improved in the CKD stage, with 45% of the participants improving from stage 3A to stage 2. GFR improved to 15 mL/min/1.73m2 (p = 0.025), and there was a negative correlation (rs = -0.3556) between weight and GFR (p = 0.013). Participants with hyperlipidemia had a 12 mL/min/1.73m2 rise in GFR, while participants without the diagnosis at one year had a 24 mL/min/1.73m2 rise in GFR (p = 0.007).ConclusionsThis study demonstrated improved renal function and reduced progression of CKD following a combined lifestyle and surgical intervention, indicating the importance of a comprehensive approach for the management of the chronic disease.

Reconstruction of the Radar Reflectivity of Convective Storms Based on Deep Learning and Himawari-8 Observations

Radar reflectivity (RR) greater than 35 dBZ usually indicates the presence of severe convective weather, which affects a variety of human activities, including aviation. However, RR data are scarce, especially in regions with poor radar coverage or substantial terrain obstructions. Fortunately, the radiance data of space-based satellites with universal coverage can be converted into a proxy field of RR. In this study, a convolutional neural network-based data-driven model is developed to convert the radiance data (infrared bands 07, 09, 13, 16, and 16–13) of Himawari-8 into the radar combined reflectivity factor (CREF). A weighted loss function is designed to solve the data imbalance problem due to the sparse convective pixels in the sample. The developed model demonstrates an overall reconstruction capability and performs well in terms of classification scores with 35 dBZ as the threshold. A five-channel input is more efficient in reconstructing the CREF than the commonly used one-channel input. In a case study of a convective event over North China in the summer using the test dataset, U-Net reproduces the location, shape and strength of the convective storm well. The present RR reconstruction technology based on deep learning and Himawari-8 radiance data is shown to be an efficient tool for producing high-resolution RR products, which are especially needed for regions without or with poor radar coverage.

Correction of uneven illumination in color microscopic image based on fully convolutional network.

The correction of uneven illumination in microscopic image is a basic task in medical imaging. Most of the existing methods are designed for monochrome images. An effective fully convolutional network (FCN) is proposed to directly process color microscopic image in this paper. The proposed method estimates the distribution of illumination information in input image, and then carry out the correction of the corresponding uneven illumination through a feature encoder module, a feature decoder module, and a detail supplement module. In this process, overlapping residual blocks are designed to better transfer the illumination information, and in particular a well-designed weighted loss function ensures that the network can not only correct the illumination but also preserve image details. The proposed method is compared with some related methods on real pathological cell images qualitatively and quantitatively. Experimental results show that our method achieves the excellent performance. The proposed method is also applied to the preprocessing of whole slide imaging (WSI) tiles, which greatly improves the effect of image mosaicking.

Is simultaneous panniculectomy an ideal approach to repair a ventral hernia: a general surgeon’s experience

The number of Australian patients undergoing ventral hernia repair has seen a significant increase in the last twenty years. With the obesity epidemic, the general surgeon is frequently seeing patients with hernias and significant abdominal aprons and is being asked to address this at the time of the hernia repair. This can be performed utilising a general surgery and plastic surgery team, but there may be some advantages to general surgeon being able incorporate this into their practice. We present our approach to patients undergoing ventral hernia repair and simultaneous panniculectomy (VHR + PAN) by a single general surgeon. Data were analysed from a single surgeon's experience performing VHR + PAN at the same operation. Data were collected prospectively from 2009 to 2020. 146 cases of patients undergoing VHR + PAN were identified and included in this study. The mean age of patients undergoing VHR + PAN was 58years. The mean BMI was 35, with 59% of patients losing weight loss prior to surgery. 66% of patients had a hernia repair with biosynthetic mesh and 91% of patients had retro-rectus or pre-peritoneal mesh approach to the repair. 42% of patients had a post-operative complication with 80.6% of these being related to surgical-site occurrences. Other complications included gastrointestinal (14%), respiratory (13%) and venous thromboembolism, such as a deep vein thrombus or pulmonary embolism (6%). There were 2 deaths in the series (1.3%). The hernia recurrence rate was 6%. Simultaneous PAN is possible in patients with an abdominal apron who are undergoing VHR, with an acceptable risk of SSOs and other complications. This technique provides excellent exposure and with appropriate training is well within the remit of the general surgeon. This may save further operative management in the future and can offer patients improved self-esteem, mobility, and independence. Patient optimisation is key, paying careful attention to pre-operative weight loss, diabetic control, smoking cessation and respiratory function. VHR + PAN is an important technique that should be in the repertoire of all abdominal wall reconstruction units.

Beta Cell Function as a Baseline Predictor of Weight Loss After Bariatric Surgery.

BackgroundObesity is a multifactorial disease, which is strongly associated to other metabolic disorders. Bariatric surgery is the most effective treatment of morbid obesity. The role of beta cell function in weight loss after bariatric surgery is uncertain.AimTo evaluate the association between beta cell function and percentage of total body weight loss (TBWL%) 1, 2, 3, and 4 years after bariatric surgery in patients with morbid obesity.MethodsRetrospective longitudinal study in patients with morbid obesity followed in our center between January 2010 and July 2018. Patients were excluded if they had diabetes at baseline or missing data on the needed parameters. We evaluated baseline Homeostatic Model Assessment of IR, Homeostatic Model Assessment of β-cell function (HOMA-beta), Quantitative Insulin Sensitivity Check Index, and Matsuda and DeFronzo index, and TBWL% at years 1 to 4. Linear regression models were used to evaluate the association of indexes of insulin resistance with TBWL% (unadjusted and adjusted for age, sex, BMI, and type of surgery).ResultsThere were 1,561 patients included in this analysis. HOMA-beta was negatively associated with TBWL% at second, third, and fourth years post-surgery (β = −1.04 [−1.82 to −0.26], p<0.01; β = −1.16 [−2.13 to −0.19], p=0.02; β = −1.29 [−2.64 to 0.06], p=0.061, respectively). This was not observed in the first year post-surgery nor for the other indexes. Glycemia at baseline was positively associated to EWL% at second and third years post-surgery.Conclusionβ-cell function at baseline seems to be associated to long-term weight loss, explicitly after the first year post bariatric surgery. This might be a helpful predictor of weight loss in clinical practice.

Group cognitive remediation therapy prior to behavioral weight loss treatment for adults with severe obesity: A randomized clinical trial (CRT study).

Objective: Individuals with obesity show executive dysfunctions that have been implicated in weight management failure. Initial evidence suggests that cognitive remediation therapy (CRT) conducted after behavioral weight loss (BWL) treatment improves weight loss and executive function, but efficacy for CRT conducted before BWL treatment is unknown. This study investigated whether group CRT in adults with Class II or III obesity (body mass index, BMI≥35 kg/m2) improves weight loss, executive function, weight management behavior, and mental and physical health in real-world group BWL treatment. Method: In this prospective single-center, assessor-blind trial (DRKS00009333), 270 adults with Class II and III obesity (age 44.5 ± 12.8 years, BMI 45.6 ± 6.9 kg/m2, and 68.9% women) were randomized to CRT with 8 group sessions over 2 months versus no treatment control, followed by routine BWL treatment of up to 12 months for both groups. The primary outcome was percent weight change at 6 months. Secondary outcomes included executive functions, weight management behaviors, and mental and physical health. Results: In intent-to-treat analyses, overall weight loss after 6 months was 1.2%, 95% CI [-2.0% to -0.4%], p = .002. The difference between arms was 0.4%, 95% CI [-1.1% to 1.8%], p = .629, Cohen's d = 0.09, after 6 months and 0.3%, 95% CI [-1.5% to 2.2%], p = .721, Cohen's d = 0.01, after 12 months. Improvements in most secondary outcomes including executive functions were seen at most time points, however, without differences between arms. Conclusions: Group CRT versus no treatment prior to real-world BWL treatment in adults with Class II and III obesity does not improve weight loss. (PsycInfo Database Record (c) 2021 APA, all rights reserved).

A deep embedded refined clustering approach for breast cancer distinction based on DNA methylation

Epigenetic alterations have an important role in the development of several types of cancer. Epigenetic studies generate a large amount of data, which makes it essential to develop novel models capable of dealing with large-scale data. In this work, we propose a deep embedded refined clustering method for breast cancer differentiation based on DNA methylation. In concrete, the deep learning system presented here uses the levels of CpG island methylation between 0 and 1. The proposed approach is composed of two main stages. The first stage consists in the dimensionality reduction of the methylation data based on an autoencoder. The second stage is a clustering algorithm based on the soft assignment of the latent space provided by the autoencoder. The whole method is optimized through a weighted loss function composed of two terms: reconstruction and classification terms. To the best of the authors’ knowledge, no previous studies have focused on the dimensionality reduction algorithms linked to classification trained end-to-end for DNA methylation analysis. The proposed method achieves an unsupervised clustering accuracy of 0.9927 and an error rate (%) of 0.73 on 137 breast tissue samples. After a second test of the deep-learning-based method using a different methylation database, an accuracy of 0.9343 and an error rate (%) of 6.57 on 45 breast tissue samples are obtained. Based on these results, the proposed algorithm outperforms other state-of-the-art methods evaluated under the same conditions for breast cancer classification based on DNA methylation data.

Building Multi-Feature Fusion Refined Network for Building Extraction from High-Resolution Remote Sensing Images

Deep learning approaches have been widely used in building automatic extraction tasks and have made great progress in recent years. However, the missing detection and wrong detection causing by spectrum confusion is still a great challenge. The existing fully convolutional networks (FCNs) cannot effectively distinguish whether the feature differences are from one building or the building and its adjacent non-building objects. In order to overcome the limitations, a building multi-feature fusion refined network (BMFR-Net) was presented in this paper to extract buildings accurately and completely. BMFR-Net is based on an encoding and decoding structure, mainly consisting of two parts: the continuous atrous convolution pyramid (CACP) module and the multiscale output fusion constraint (MOFC) structure. The CACP module is positioned at the end of the contracting path and it effectively minimizes the loss of effective information in multiscale feature extraction and fusion by using parallel continuous small-scale atrous convolution. To improve the ability to aggregate semantic information from the context, the MOFC structure performs predictive output at each stage of the expanding path and integrates the results into the network. Furthermore, the multilevel joint weighted loss function effectively updates parameters well away from the output layer, enhancing the learning capacity of the network for low-level abstract features. The experimental results demonstrate that the proposed BMFR-Net outperforms the other five state-of-the-art approaches in both visual interpretation and quantitative evaluation.

Nox4 Maintains Blood Pressure during Low Sodium Diet.

The NADPH oxidase Nox4 is a hydrogen peroxide (H2O2)-producing enzyme, with the highest expression in the kidney. As the kidney is involved in volume and blood pressure control through sodium handling, we set out to determine the impact of a low sodium diet on these parameters in WT and Nox4-/- mice. Nox4 expression in the murine kidney was restricted to the proximal tubule. Nevertheless, low-sodium-induced weight loss and sodium sparing function was similar in WT and Nox4-/- mice, disputing an important function of renal Nox4 in sodium handling. In contrast, a low sodium diet resulted in a reduction in systolic blood pressure in Nox4-/- as compared to WT mice. This was associated with a selectively lower pressure to heart-rate ratio, as well as heart to body weight ratio. In general, a low sodium diet leads to activation of sympathetic tone and the renin angiotensin system, which subsequently increases peripheral resistance. Our observations suggest that the control by this system is attenuated in Nox4-/- mice, resulting in lower blood pressure in response to low sodium.

A 2020 perspective on “DeRec: A data-driven approach to accurate recommendation with deep learning and weighted loss function”

The development of Internet comes up with the prosperity of E-commerce all over the world. In order to promote sales and save consumers’ labor in commodity browsing, recommender systems are proposed by E-commerce platforms to provide online consumers with products and services of their potential interests. The primary challenge in recommendation roots in the intricacy in quantifying users’ preferences on items with the reality of data sparsity and the computation complexity. Hence, more and more researchers are attempting deep learning techniques to deal with the challenge with the hope of using advanced algorithms to alleviate the intricacy. Word embedding is used to learn the association of items in a space of low dimensionality. Multi-layer perception is used to learn users’ preferences on items in a data-driven manner with a customized loss function. The future work of recommender systems includes three folds. The one is to make use of multi-source data to combine implicit and explicit user behavior data to address the problem of data sparsity. The second is dynamic recommendation with the changing users’ preferences on items and make recommender systems light-weight and useable in complex scenarios. The third is to provide effective and verifiable recommendation under the premise of user privacy protection

Bayesian Estimation of Stress-Strength P(T < X < Z) for Dagum Distribution

In this paper, the reliability formula is derived for the stress-strength model of the probability P(T < X < Z) for a component’s strength X falling between two stresses T and Z, based on Dagum Distribution with unknown parameter β and known and common parameters λ and δ. Bayesian estimation is discussed to estimate the reliability under complete data by using Gamma prior based on two loss function (weighted and quadratic loss functions), and the comparison between these estimators based on a simulation study using mean square error criteria (MSE) for each of the small, medium and large samples. The most important conclusion is that this comparison confirms that the performance of the estimator according to the weighted loss function works better for the most experiments studied.

Comparing Weibull Stress – Strength Reliability Bayesian Estimators for Singly Type II Censored Data under Different loss Functions

The stress(Y) – strength(X) model reliability Bayesian estimation which defines life of a component with strength X and stress Y (the component fails if and only if at any time the applied stress is greater than its strength) has been studied, then the reliability; R=P(Y&lt;X), can be considered as a measure of the component performance. In this paper, a Bayesian analysis has been considered for R when the two variables X and Y are independent Weibull random variables with common parameter α in order to study the effect of each of the two different scale parameters β and λ; respectively, using three different [weighted, quadratic and entropy] loss functions under two different prior functions [Gamma and extension of Jeffery] and also an empirical Bayes estimator Using Gamma Prior, for singly type II censored sample. An empirical study has been used to make a comparison between the three estimators of the reliability for stress – strength Weibull model, by mean squared error MSE criteria, taking different sample sizes (small, moderate and large) for the two random variables in eight experiments of different values of their parameters. It has been found that the weighted loss function was the best for small sample size, and the entropy and Quadratic were the best for moderate and large sample sizes under the two prior distributions and for empirical Bayes estimation.

Preference of Prior for Two-Component Mixture of Lomax Distribution

Recently, El-Sherpieny et al., (2020), suggested Type-II hybrid censoring method for parametric estimation of Lomax distribution (LD) without due regard being given to the choice of priors and posterior risk associated with the model. This paper fills this gap and derived the new LD model with minimum posterior risk for the selection of priors. It derives a closed form expression for Bayes estimates and posterior risks using square error loss function (SELF), weighted loss function (WLF), quadratic loss function (QLF) and DeGroot loss function (DLF). Prior predictive approach is used to elicit the hyperparameters of mixture model. Analysis of Bayes estimates and posterior risks is presented in terms of sample size (n), mixing proportion (p) and censoring rate (t0), with the help of simulation study. Usefulness of the model is demonstrated on applying it to simulated and real-life data which show promising results in terms of better estimation and risk reduction.

Adaptive Weighted Discriminator for Training Generative Adversarial Networks.

Generative adversarial network (GAN) has become one of the most important neural network models for classical unsupervised machine learning. A variety of discriminator loss functions have been developed to train GAN's discriminators and they all have a common structure: a sum of real and fake losses that only depends on the actual and generated data respectively. One challenge associated with an equally weighted sum of two losses is that the training may benefit one loss but harm the other, which we show causes instability and mode collapse. In this paper, we introduce a new family of discriminator loss functions that adopts a weighted sum of real and fake parts, which we call adaptive weighted loss functions or aw-loss functions. Using the gradients of the real and fake parts of the loss, we can adaptively choose weights to train a discriminator in the direction that benefits the GAN's stability. Our method can be potentially applied to any discriminator model with a loss that is a sum of the real and fake parts. For our experiments, SN-GAN, AutoGAN, and BigGAN are used. Experiments validated the effectiveness of our loss functions on unconditional and conditional image generation tasks, improving the baseline results by a significant margin on CIFAR-10, STL-10, and CIFAR-100 datasets in Inception Scores (IS) and Fréchet Inception Distance (FID) metrics.