Search

Pneumonia has long been a significant concern in global public health. With the advancement of convolutional neural networks (CNNs), new technological methods have emerged to address this challenge. However, the application of CNNs to pneumonia diagnosis still faces several critical issues. First, the datasets used for training models often suffer from insufficient sample sizes and imbalanced class distributions, leading to reduced classification performance. Second, although CNNs can automatically extract features and make decisions from complex image data, their interpretability is relatively poor, limiting their widespread use in clinical diagnosis to some extent. To address these issues, a novel weighted cross-entropy loss function is proposed, which calculates weights via an inverse proportion exponential function to handle data imbalance more efficiently. Additionally, we employ a transfer learning approach that combines pretrained CNN model parameter fine-tuning to improve classification performance. Finally, we introduce the gradient-weighted class activation mapping method to enhance the interpretability of the model's decisions by visualizing the image regions of focus. The experimental results indicate that our proposed approach significantly enhances CNN performance in pneumonia diagnosis tasks. Among the four selected models, the accuracy rates improved to over 90%, and visualized results were provided. [ABSTRACT FROM AUTHOR]

Simple Summary: Bladder cancer segmentation on MRI images is critical to determine if the cancer spread to the nearby muscles. In this study, we aimed to assess the performance of three deep learning models in outlining bladder tumors from MRI images. Using the MRI data of 53 patients, we trained Unet, MAnet, and PSPnet models to segment tumors using different loss functions and evaluated their performances. The results showed MAnet and PSPnet models performed better overall in segmenting bladder tumors, especially when they used a hybrid loss function (CE+DSC). Our findings could improve the way bladder cancer is segmented on MRI images, potentially leading to a better choice of deep learning algorithms and loss functions for future research. Background: Bladder cancer (BC) segmentation on MRI images is the first step to determining the presence of muscular invasion. This study aimed to assess the tumor segmentation performance of three deep learning (DL) models on multi-parametric MRI (mp-MRI) images. Methods: We studied 53 patients with bladder cancer. Bladder tumors were segmented on each slice of T2-weighted (T2WI), diffusion-weighted imaging/apparent diffusion coefficient (DWI/ADC), and T1-weighted contrast-enhanced (T1WI) images acquired at a 3Tesla MRI scanner. We trained Unet, MAnet, and PSPnet using three loss functions: cross-entropy (CE), dice similarity coefficient loss (DSC), and focal loss (FL). We evaluated the model performances using DSC, Hausdorff distance (HD), and expected calibration error (ECE). Results: The MAnet algorithm with the CE+DSC loss function gave the highest DSC values on the ADC, T2WI, and T1WI images. PSPnet with CE+DSC obtained the smallest HDs on the ADC, T2WI, and T1WI images. The segmentation accuracy overall was better on the ADC and T1WI than on the T2WI. The ECEs were the smallest for PSPnet with FL on the ADC images, while they were the smallest for MAnet with CE+DSC on the T2WI and T1WI. Conclusions: Compared to Unet, MAnet and PSPnet with a hybrid CE+DSC loss function displayed better performances in BC segmentation depending on the choice of the evaluation metric. [ABSTRACT FROM AUTHOR]

Non-Euclidean data, such as social networks and citation relationships between documents, have node and structural information. The Graph Convolutional Network (GCN) can automatically learn node features and association information between nodes. The core ideology of the Graph Convolutional Network is to aggregate node information by using edge information, thereby generating a new node feature. In updating node features, there are two core influencing factors. One is the number of neighboring nodes of the central node; the other is the contribution of the neighboring nodes to the central node. Due to the previous GCN methods not simultaneously considering the numbers and different contributions of neighboring nodes to the central node, we design the adaptive attention mechanism (AAM). To further enhance the representational capability of the model, we utilize Multi-Head Graph Convolution (MHGC). Finally, we adopt the cross-entropy (CE) loss function to describe the difference between the predicted results of node categories and the ground truth (GT). Combined with backpropagation, this ultimately achieves accurate node classification. Based on the AAM, MHGC, and CE, we contrive the novel Graph Adaptive Attention Network (GAAN). The experiments show that classification accuracy achieves outstanding performances on Cora, Citeseer, and Pubmed datasets. [ABSTRACT FROM AUTHOR]

The existing segmentation-based scene text detection methods mostly need complicated post-processing, and the post-processing operation is separated from the training process, which greatly reduces the detection performance. The previous method, DBNet, successfully simplified post-processing and integrated post-processing into a segmentation network. However, the training process of the model took a long time for 1200 epochs and the sensitivity to texts of various scales was lacking, leading to some text instances being missed. Considering the above two problems, we design the text detection Network with Binarization of Hyperbolic Tangent (HTBNet). First of all, we propose the Binarization of Hyperbolic Tangent (HTB), optimized along with which the segmentation network can expedite the initial convergent speed by reducing the number of epochs from 1200 to 600. Because features of different channels in the same scale feature map focus on the information of different regions in the image, to better represent the important features of all objects in the image, we devise the Multi-Scale Channel Attention (MSCA). Meanwhile, considering that multi-scale objects in the image cannot be simultaneously detected, we propose a novel module named Fused Module with Channel and Spatial (FMCS), which can fuse the multi-scale feature maps from channel and spatial dimensions. Finally, we adopt cross-entropy as the loss function, which measures the difference between predicted values and ground truths. The experimental results show that HTBNet, compared with lightweight models, has achieved competitive performance and speed on Total-Text (F-measure:86.0%, FPS:30) and MSRA-TD500 (F-measure:87.5%, FPS:30). [ABSTRACT FROM AUTHOR]

Estimation of dynamic mixture distributions is a difficult task, because the density contains an intractable normalizing constant. To overcome this difficulty, we develop an approach that maximizes, by means of the cross‐entropy method, a Monte Carlo approximation of the log‐likelihood function. The proposed noisy cross‐entropy approach is unsupervised, since it does not require the specification of a threshold between the distributions. Moreover, it bypasses the evaluation of the normalizing constant, combining good statistical properties with a modest computational burden. Both simulation‐based evidence and empirical applications suggest that noisy cross‐entropy estimation is comparable or preferable to existing methods in terms of statistical efficiency, but is less demanding from the computational point of view. [ABSTRACT FROM AUTHOR]

Multi-criteria decision-making (MCDM) entails a heterogeneous decision-making problem, which poses challenges for decision-makers (DMs) in generating an optimal solution. To address this, we have proposed a heterogeneous evaluation framework. First, a novel picture fuzzy cross-entropy measure was defined with the simultaneous consideration of uncertainty and hesitancy of picture fuzzy information, overcoming the shortcomings of the existing cross-entropy measure in relation to its validity and properties. Next, an optimisation-model for determining the objective weights of criteria was constructed based on the proposed closeness measurement and the step-wise weight assessment ratio analysis (SWARA) method. This model was constituted from both objective and subjective perspectives under the circumstance of completely unknown criterion information. Additionally, the conventional Measurement of Alternatives and Ranking according to Compromise Solution (MARCOS) method was extended to the picture fuzzy environment, and a normalisation method was developed to transform heterogeneous judgments, including linguistic terms, interval numbers, and picture fuzzy numbers, into a unified representation form. A heterogeneous picture fuzzy SWARA-MARCOS evaluation framework was then established and used to solve a military equipment supplier selection problem. The results demonstrated the validity and feasibility of the proposed evaluation framework, while sensitivity, comparative, and complexity analyses demonstrated the robustness and superiority of it. [ABSTRACT FROM AUTHOR]

Cross-entropy loss is crucial in training many deep neural networks. In this context, we show a number of novel and strong correlations among various related divergence functions. In particular, we demonstrate that, in some circumstances, (a) cross-entropy is almost perfectly correlated with the little-known triangular divergence, and (b) cross-entropy is strongly correlated with the Euclidean distance over the logits from which the softmax is derived. The consequences of these observations are as follows. First, triangular divergence may be used as a cheaper alternative to cross-entropy. Second, logits can be used as features in a Euclidean space which is strongly synergistic with the classification process. This justifies the use of Euclidean distance over logits as a measure of similarity, in cases where the network is trained using softmax and cross-entropy. We establish these correlations via empirical observation, supported by a mathematical explanation encompassing a number of strongly related divergence functions. [ABSTRACT FROM AUTHOR]

As an effective tool to show the fuzziness of qualitative information, the interval-valued T-spherical fuzzy set can utilize three kinds of information, namely, membership, abstinence, and non-membership, to show the opinions of decision-maker. Given this advantage, many interval-valued T-spherical fuzzy multi-criteria decision-making (IVTSF-MCDM) methods have been designed. However, most of the existing IVTSF-MCDM methods have a common limitation that the inability to effectively show the impacts of extreme data. To address this limitation, this study develops a novel MCDM method based on interval-valued T-spherical fuzzy extended power aggregation operator. First, interval-valued T-spherical fuzzy cross-entropy (CE) and interval-valued T-spherical fuzzy symmetrical CE are defined to measure the difference between two interval-valued T-spherical fuzzy numbers, which are used to determine criteria weights in MCDM. Second, interval-valued T-spherical fuzzy extended power average operator and interval-valued T-spherical fuzzy extended power geometric operator are proposed, and their properties are investigated. Moreover, in view of that criteria may be assigned to different weights, this study defines interval-valued T-spherical fuzzy extended power weighted average operator and interval-valued T-spherical fuzzy extended power weighted geometric operator to derive the order of alternatives. Finally, the applicability of the proposed method is validated by the case about investment country selection, while the sensitivity and comparison analyses are also conducted to further prove its advantages and effectiveness.

Autism spectrum disorder (ASD) is an enduring and intricate condition that encompasses issues with behavior and communication. This disability has become common among several individuals globally. To ensure that all the autistic individuals get the correct support and care throughout their lifetime, it is significant that the autism signs are realized and the condition is spotted in the initial phase itself to avoid further complications. However, the process of ASD diagnosis is time-consuming, with expensive testing procedures. An effective screening method is needed for early treatment to improve the quality of the ASD patient's life. Several conventional studies thrived on attaining ASD identification with machine learning and deep learning) but lacked in accuracy and computation. Therefore, the propounded system employs modified multi-layer perceptron (MLP) with cross-weighted attention mechanism. MLP is utilized for its capability in resolving intricate nonlinear complications, handling huge datasets, and enhancing the accuracy of the model. Though existing studies have utilized MLP for various identification processes, it still lacks identification speed, has overfitting issues, and requires more parameters. To overcome these issues, the proposed system employs cross-weighted attention mechanism, to enhance the identification process. As few researchers focused only on identifying and diagnosing the ASD adult, the proposed system uses autism screening in adult datasets and intends to improve the diagnosis of ASD adult patients. Attention ASD-modified MLP with cross-weighted algorithm is applied to classify and perform with various algorithms such as random forest, MLP, and Naïve Bayes. Furthermore, the performance is examined with certain metrics to calculate the efficacy of the proposed system. [ABSTRACT FROM AUTHOR]

A T-spherical uncertain linguistic set (TSULS) is not only an expanded form of the T-spherical fuzzy set and the uncertain linguistic set but can also integrate the quantitative judging ideas and qualitative assessing information of decision-makers. For the description of complex and uncertain assessment data, TSULS is a powerful tool for the precise description and reliable processing of information data. However, the existing multi-attribute border approximation area comparison (MABAC) method has not been studied in TSULS. Thus, the goal of this paper is to extend and improve the MABAC method to tackle group decision-making problems with completely unknown weight information in the TSUL context. First, the cross-entropy measure and the interactive operation laws for the TSUL numbers are defined, respectively. Then, the two interactive aggregation operators for TSUL numbers are developed, namely T-spherical uncertain linguistic interactive weighted averaging and T-spherical uncertain linguistic interactive weighted geometric operators. Their effective properties and some special cases are also investigated. Subsequently, a new TSULMAGDM model considering the DM's behavioral preference and psychology is built by integrating the interactive aggregation operators, the cross-entropy measure, prospect theory, and the MABAC method. To explore the effectiveness and practicability of the proposed model, an illustrative example of Sustainable Waste Clothing Recycling Partner selection is presented, and the results show that the optimal solution is h3. Finally, the reliable, valid, and generalized nature of the method is further verified through sensitivity analysis and comparative studies with existing methods. [ABSTRACT FROM AUTHOR]

Joint models for longitudinal and time‐to‐event data are often employed to calculate dynamic individualized predictions used in numerous applications of precision medicine. Two components of joint models that influence the accuracy of these predictions are the shape of the longitudinal trajectories and the functional form linking the longitudinal outcome history to the hazard of the event. Finding a single well‐specified model that produces accurate predictions for all subjects and follow‐up times can be challenging, especially when considering multiple longitudinal outcomes. In this work, we use the concept of super learning and avoid selecting a single model. In particular, we specify a weighted combination of the dynamic predictions calculated from a library of joint models with different specifications. The weights are selected to optimize a predictive accuracy metric using V‐fold cross‐validation. We use as predictive accuracy measures the expected quadratic prediction error and the expected predictive cross‐entropy. In a simulation study, we found that the super learning approach produces results very similar to the Oracle model, which was the model with the best performance in the test datasets. All proposed methodology is implemented in the freely available R package JMbayes2. [ABSTRACT FROM AUTHOR]

Steel surfaces often display intricate texture patterns that can resemble defects, posing a challenge in accurately identifying actual defects. Therefore, it is crucial to develop a highly robust defect detection model. This study proposes a defect detection method for steel infrared images based on a Regularized YOLO framework. Firstly, the Coordinate Attention (CA) is embedded within the C2F framework, utilizing a lightweight attention module to enhance the feature extraction capability of the backbone network. Secondly, the neck part design incorporates the Bi-directional Feature Pyramid Network (BiFPN) for weighted fusion of multi-scale feature maps. This creates a model called BiFPN-Concat, which enhances feature fusion capability. Finally, the loss function of the model is regularized to improve the generalization performance of the model. The experimental results indicate that the model has only 3.03 M parameters, yet achieves a mAP@0.5 of 80.77% on the NEU-DET dataset and 99.38% on the ECTI dataset. This represents an improvement of 2.3% and 1.6% over the baseline model, respectively. This method is well-suited for industrial detection applications involving non-destructive testing of steel using infrared imagery. [ABSTRACT FROM AUTHOR]

The purpose of the study is to propose a multi-class ensemble classifier using interval modeling dedicated to microarray datasets. An approach of creating the uncertainty intervals for the single prediction values of constituent classifiers and then aggregating the obtained intervals with the use of interval-valued aggregation functions is used. The proposed heterogeneous classification employs Random Forest, Support Vector Machines, and Multilayer Perceptron as component classifiers, utilizing cross-entropy to select the optimal classifier. Moreover, orders for intervals are applied to determine the decision class of an object. The applied interval-valued aggregation functions are tested in terms of optimizing the performance of the considered ensemble classifier. The proposed model's quality, superior to other well-known and component classifiers, is validated through comparison, demonstrating the efficacy of cross-entropy in ensemble model construction. [ABSTRACT FROM AUTHOR]

Abstract A T-spherical uncertain linguistic set (TSULS) is not only an expanded form of the T-spherical fuzzy set and the uncertain linguistic set but can also integrate the quantitative judging ideas and qualitative assessing information of decision-makers. For the description of complex and uncertain assessment data, TSULS is a powerful tool for the precise description and reliable processing of information data. However, the existing multi-attribute border approximation area comparison (MABAC) method has not been studied in TSULS. Thus, the goal of this paper is to extend and improve the MABAC method to tackle group decision-making problems with completely unknown weight information in the TSUL context. First, the cross-entropy measure and the interactive operation laws for the TSUL numbers are defined, respectively. Then, the two interactive aggregation operators for TSUL numbers are developed, namely T-spherical uncertain linguistic interactive weighted averaging and T-spherical uncertain linguistic interactive weighted geometric operators. Their effective properties and some special cases are also investigated. Subsequently, a new TSULMAGDM model considering the DM’s behavioral preference and psychology is built by integrating the interactive aggregation operators, the cross-entropy measure, prospect theory, and the MABAC method. To explore the effectiveness and practicability of the proposed model, an illustrative example of Sustainable Waste Clothing Recycling Partner selection is presented, and the results show that the optimal solution is h 3. Finally, the reliable, valid, and generalized nature of the method is further verified through sensitivity analysis and comparative studies with existing methods.

Siamese network-based visual tracking algorithms have achieved excellent performance in recent years, but challenges such as fast target motion, shape and scale variations have made the tracking extremely difficult. The regression of anchor-free tracking has low computational complexity, strong real-time performance, and is suitable for visual tracking. Based on the anchor-free siamese tracking framework, this paper firstly introduces balance factors and modulation coefficients into the cross-entropy loss function to solve the classification inaccuracy caused by the imbalance between positive and negative samples as well as the imbalance between hard and easy samples during the training process, so that the model focuses more on the positive samples and the hard samples that make the major contribution to the training. Secondly, the intersection over union (IoU) loss function of the regression branch is improved, not only focusing on the IoU between the predicted box and the ground truth box, but also considering the aspect ratios of the two boxes and the minimum bounding box area that accommodate the two, which guides the generation of more accurate regression offsets. The overall loss of classification and regression is iteratively minimized and improves the accuracy and robustness of visual tracking. Experiments on four public datasets, OTB2015, VOT2016, UAV123 and GOT-10k, show that the proposed algorithm achieves the state-of-the-art performance.

Question matching is the fundamental task in retrieval-based dialogue systems which assesses the similarity between Query and Question. Unfortunately, existing methods focus on improving the accuracy of text similarity in the general domain, without adaptation to the financial domain. Financial question matching has two critical issues: (1) How to accurately model the contextual representation of a financial sentence? (2) How to accurately represent financial key phrases in an utterance? To address these issues, this paper proposes a novel Financial Knowledge Enhanced Network (FinKENet) that significantly injects financial knowledge into contextual text. Specifically, we propose a multi-level encoder to extract both sentence-level features and financial phrase-level features, which can more accurately represent sentences and financial phrases. Furthermore, we propose a financial co-attention adapter to combine sentence features and financial keyword features. Finally, we design a multi-level similarity decoder to calculate the similarity between queries and questions. In addition, a cross-entropy-based loss function is presented for model optimization. Experimental results demonstrate the effectiveness of the proposed method on the Ant Financial question matching dataset. In particular, the Recall score improves from 73.21% to 74.90% (1.69% absolute). [ABSTRACT FROM AUTHOR]

Non-Euclidean data, such as social networks and citation relationships between documents, have node and structural information. The Graph Convolutional Network (GCN) can automatically learn node features and association information between nodes. The core ideology of the Graph Convolutional Network is to aggregate node information by using edge information, thereby generating a new node feature. In updating node features, there are two core influencing factors. One is the number of neighboring nodes of the central node; the other is the contribution of the neighboring nodes to the central node. Due to the previous GCN methods not simultaneously considering the numbers and different contributions of neighboring nodes to the central node, we design the adaptive attention mechanism (AAM). To further enhance the representational capability of the model, we utilize Multi-Head Graph Convolution (MHGC). Finally, we adopt the cross-entropy (CE) loss function to describe the difference between the predicted results of node categories and the ground truth (GT). Combined with backpropagation, this ultimately achieves accurate node classification. Based on the AAM, MHGC, and CE, we contrive the novel Graph Adaptive Attention Network (GAAN). The experiments show that classification accuracy achieves outstanding performances on Cora, Citeseer, and Pubmed datasets.

The existing segmentation-based scene text detection methods mostly need complicated post-processing, and the post-processing operation is separated from the training process, which greatly reduces the detection performance. The previous method, DBNet, successfully simplified post-processing and integrated post-processing into a segmentation network. However, the training process of the model took a long time for 1200 epochs and the sensitivity to texts of various scales was lacking, leading to some text instances being missed. Considering the above two problems, we design the text detection Network with Binarization of Hyperbolic Tangent (HTBNet). First of all, we propose the Binarization of Hyperbolic Tangent (HTB), optimized along with which the segmentation network can expedite the initial convergent speed by reducing the number of epochs from 1200 to 600. Because features of different channels in the same scale feature map focus on the information of different regions in the image, to better represent the important features of all objects in the image, we devise the Multi-Scale Channel Attention (MSCA). Meanwhile, considering that multi-scale objects in the image cannot be simultaneously detected, we propose a novel module named Fused Module with Channel and Spatial (FMCS), which can fuse the multi-scale feature maps from channel and spatial dimensions. Finally, we adopt cross-entropy as the loss function, which measures the difference between predicted values and ground truths. The experimental results show that HTBNet, compared with lightweight models, has achieved competitive performance and speed on Total-Text (F-measure:86.0%, FPS:30) and MSRA-TD500 (F-measure:87.5%, FPS:30).

Cross-entropy loss is crucial in training many deep neural networks. In this context, we show a number of novel and strong correlations among various related divergence functions. In particular, we demonstrate that, in some circumstances, (a) cross-entropy is almost perfectly correlated with the little-known triangular divergence, and (b) cross-entropy is strongly correlated with the Euclidean distance over the logits from which the softmax is derived. The consequences of these observations are as follows. First, triangular divergence may be used as a cheaper alternative to cross-entropy. Second, logits can be used as features in a Euclidean space which is strongly synergistic with the classification process. This justifies the use of Euclidean distance over logits as a measure of similarity, in cases where the network is trained using softmax and cross-entropy. We establish these correlations via empirical observation, supported by a mathematical explanation encompassing a number of strongly related divergence functions.

In this paper, a multi-branch deep network architecture (JESF-MBF) based on joint embedded semantic features is proposed. In order to construct the structural consistency between visual image space and label semantic space, as well as to ensure the correspondence between sketch or image data and labels, the JESF-MBF model uses cross-entropy-based multivariate classification loss and cosine distance-based semantic similarity loss for model optimization. The optimized model improves the ability of label semantic information representation and image visual information representation, alleviates the discrepancy of data domain distribution between sketch and image, and constructs the mapping relationship between visual embedding space and semantic embedding space, so the sketch-based zero-sample image retrieval task is realized based on the intermediate medium. In the comparative model analysis, the average evaluation index of JESF-MBF is 0.412, better than that of Doodle2search at 0.344. In terms of the teaching effectiveness of the teaching platform constructed in this paper, the average time spent for the experimental class to complete each subject work is 9.4 class periods, while the average time required for the experimental class is 19.7 class periods. The results show that the art teaching platform based on the JESF-MBF model is very effective in terms of performance as well as improving the art level.

Multilevel thresholding is one of the most commonly used methods in image segmentation. However, the exhaustive search methods are costly in determining optimal thresholds and the conventional remora optimization algorithm (ROA) is prone to the premature convergence. This paper presents a chimp-inspired remora optimization algorithm (HCROA) to search optimal threshold levels, and the cross-entropy is employed as the objective function. In HCROA, the particles' position are adjusted by the Chimp Optimization Algorithm (ChOA) because of its good exploitation ability and sufficient diversity. With this change, HCROA achieves both the intra-group diversity intelligence and a suitable balance between exploration and exploitation. To validate its performance, a series of experiments are performed. First, we test the HCROA's segmentation accuracy by a set of natural gray-scale images with different thresholds. Second, HCROA is implemented for noisy image segmentation to evaluate its robustness. Several reference-based measurements including Peak Signal-to-Noise Ratio (PSNR), Structural Similarity (SSIM), Feature Similarity (FSIM), Quality Index based on Local Variance (QILV), Haar wavelet-based Perceptual Similarity Index (HPSI), Wilcoxon test, and CPU time have been considered for evaluating the proposed method. Additionally, eight well-known predecessors are injected for parallel comparison. The comparison results prove that the suggested method outperforms the existing approaches in terms of accuracy, convergence speed, noise robustness, and efficiency. [ABSTRACT FROM AUTHOR]

The Dual Credit Policy is an important policy to promote the development of new energy vehicles unique to China. There is a lack of research that intuitively reflects the impact of the Dual Credit Policy on industrial development through an industry-based factual comparison of this policy. Based on the Taylor expansion and Cross-Entropy description, this article obtains the development regression function by the quantitative analysis of five indicators—the number of new energy vehicle-related patents, sales volume, production volume, the number of newly registered enterprises, infrastructure construction (the number of charging piles) before and after the implementation of the policy, and describes them quantitatively using the Taylor expansion to obtain the CPTI index. The CPCEI index is obtained by calculating the Cross-Entropy of the distribution of each indicator before and after policy implementation. The above two indices were compared for the growth trend and growth quantity, respectively. Finally, the following conclusions were obtained: 1. the Dual Credit Policy is more significantly promoted at the market level than the impact on the technical level; 2. although there is also incentive in infrastructure construction, it cannot fully react to the market demand; 3. the number of start-up's operating in the new energy field increases, but the overall growth trend gradually slows down and fails to significantly change the existing structure of the market. This study suggests that the government should launch a special incentive policy for charging piles, and new energy manufacturers should expand their production capacity to meet the market demand. [ABSTRACT FROM AUTHOR]

Skin cancer is one of the world's most hazardous diseases, and identification of skin cancer is more challenging. Recently Deep learning algorithms have evolved to produce outstanding outcomes in a wide range of clinical practices. But classifying different classes of skin cancer is complex due to the fine-grained variability of skin lesion regions. Based on this insight, the current work considers an early and accurate multi-class skin cancer classification issue by introducing a novel deep learning model. As typical deep learning models have fewer receptive fields, they are unable to gather global context information from wider regions, making it difficult to identify diseased areas. Furthermore, in previous studies, the segmented skin lesion region caused higher noise and worse classification accuracy. To overcome the existing problem, a novel Multi-class Dilated D-Net (MD2N) framework is introduced to segment and classify multiple classes of skin cancer screening. The encoder phase of the proposed MD2N reduces feature information losses by adopting a downsampling ratio while also distinguishing small skin lesion patches. Dilated manifold Parallel convolution effectively expands the network's receptive field and eliminates the "grid issue" that plagues ordinary dilated convolution. As a result, the model can get more rich feature information of skin lesions areas of dissimilar sizes. Further, the proposed deep learning model initiates the system to accurately segment and classify lesion regions from the input skin image obtained from International Skin Imaging Collaboration. [ABSTRACT FROM AUTHOR]

Deep neural networks have made great achievements in remote sensing image analyses; however, previous studies have shown that deep neural networks exhibit incredible vulnerability to adversarial examples, which raises concerns about regional safety and production safety. In this paper, we propose an adversarial denoising method based on latent representation guidance for remote sensing image scene classification. In the training phase, we train a variational autoencoder to reconstruct the data using only the clean dataset. At test time, we first calculate the normalized mutual information between the reconstructed image using the variational autoencoder and the reference image as denoised by a discrete cosine transform. The reconstructed image is selectively utilized according to the result of the image quality assessment. Then, the latent representation of the current image is iteratively updated according to the reconstruction loss so as to gradually eliminate the influence of adversarial noise. Because the training of the denoiser only involves clean data, the proposed method is more robust against unknown adversarial noise. Experimental results on the scene classification dataset show the effectiveness of the proposed method. Furthermore, the method achieves better robust accuracy compared with state-of-the-art adversarial defense methods in image classification tasks. [ABSTRACT FROM AUTHOR]

In fuzzy decision-making, the probabilistic linguistic term sets (PLTSs) are flexible in depicting people's linguistic evaluations. As the uncertainty measures of PLTSs, entropy and cross-entropy are the important decision-making tools. Generally speaking, the uncertainty of PLTSs can be studied from two facets, which are known as hesitancy and fuzziness. To better reflect these uncertainties, some hybrid entropy and cross-entropy measures of PLTSs are developed in this work. It is shown that the hybrid entropy measures are simple in structure, clear in physical meaning and strong in discriminating ability, while the hybrid cross-entropy measures can avoid the design flaws of existing measures and possess natural symmetry. It is also shown that the hybrid entropy and cross-entropy measures can be designed in pairs, mutually supportive and inherently unified as the uncertainty measures. In the ends, a multi-attribute decision-making (MADM) model, based on these uncertainty measures of PLTSs, is developed within the famous TOPSIS framework to select the best cloud computing platform. [ABSTRACT FROM AUTHOR]

This paper handles the new information entropy measure and divergence measure associated with intuitionistic fuzzy variables (IFVs). Based on credibility distribution and credibility measure of intuitionistic fuzzy variable, the credibility entropy formulas of discrete and continuous IFVs are proposed and some of their properties are investigated. The cross-entropy of intuitionistic fuzzy variable and its relationship with credibility entropy are then discussed. Finally, some numerical examples are given to illustrate the practicability of the presented credibility entropy and cross-entropy of intuitionistic fuzzy variable. Also, we make some comparative analysis on the credibility cross-entropy measure and some existing distance measures of IFVs in the pattern recognition problem.

The purpose of the study is to propose a multi-class ensemble classifier using interval modeling dedicated to microarray datasets. An approach of creating the uncertainty intervals for the single prediction values of constituent classifiers and then aggregating the obtained intervals with the use of interval-valued aggregation functions is used. The proposed heterogeneous classification employs Random Forest, Support Vector Machines, and Multilayer Perceptron as component classifiers, utilizing cross-entropy to select the optimal classifier. Moreover, orders for intervals are applied to determine the decision class of an object. The applied interval-valued aggregation functions are tested in terms of optimizing the performance of the considered ensemble classifier. The proposed model’s quality, superior to other well-known and component classifiers, is validated through comparison, demonstrating the efficacy of cross-entropy in ensemble model construction.

Steel surfaces often display intricate texture patterns that can resemble defects, posing a challenge in accurately identifying actual defects. Therefore, it is crucial to develop a highly robust defect detection model. This study proposes a defect detection method for steel infrared images based on a Regularized YOLO framework. Firstly, the Coordinate Attention (CA) is embedded within the C2F framework, utilizing a lightweight attention module to enhance the feature extraction capability of the backbone network. Secondly, the neck part design incorporates the Bi-directional Feature Pyramid Network (BiFPN) for weighted fusion of multi-scale feature maps. This creates a model called BiFPN-Concat, which enhances feature fusion capability. Finally, the loss function of the model is regularized to improve the generalization performance of the model. The experimental results indicate that the model has only 3.03 M parameters, yet achieves a mAP@0.5 of 80.77% on the NEU-DET dataset and 99.38% on the ECTI dataset. This represents an improvement of 2.3% and 1.6% over the baseline model, respectively. This method is well-suited for industrial detection applications involving non-destructive testing of steel using infrared imagery.

Optimal power flow (OPF) is a complex, highly nonlinear, NP-hard optimization problem, in which the goal is to determine the optimal operational parameters of a power-related system (in many cases a type of smart or micro grid) which guarantee an economic and effective power dispatch. In recent years, a number of approaches based on metaheuristics algorithms have been proposed to solve OPF problems. In this paper, we propose the use of the Cross-Entropy (CE) method as a first step depth search operator to assist population-based evolutionary methods in the framework of an OPF problem. Specifically, a new variant of the Coral Reefs Optimization with Substrate Layers algorithm boosted with CE method (CE+CRO-SL) is presented in this work. We have adopted the IEEE 57-Bus System as a test scenario which, by default, has seven thermal generators for power production for the grid. We have modified this system by replacing three thermal generators with renewable source generators, in order to consider a smart grid approach with renewable energy production. The performance of CE+CRO-SL in this particular case study scenario has been compared with that of well-known techniques such as population's methods CMA-ES and EPSO (both boosted with CE). The results obtained indicate that CE+CRO-SL showed a superior performance than the alternative techniques in terms of efficiency and accuracy. This is justified by its greater exploration capacity, since it has internally operations coming from different heuristics, thus surpassing the performance of classic methods. Moreover, in a projection analysis, the CE+CRO-SL provides a profit of millions of dollars per month in all cases tested considering the modified version of the IEEE 57-Bus smart grid system. [ABSTRACT FROM AUTHOR]

The universal typical-signal estimators of entropy and cross-entropy based on the asymptotics of recurrence and waiting times play an important role in information theory. Building on their construction, we introduce and study universal typical-signal estimators of entropy production in the context of nonequilibrium statistical mechanics of one-sided shifts over finite alphabets. [ABSTRACT FROM AUTHOR]

Operational optimal control (OOC) aims to maintain the process indexes of an industrial process within the desired range by deciding the control set points. However, the optimization is made difficulty because the key process index cannot be measured online and accurate model for optimization is not available. This work deals with these issues in an industrial multiple-effect evaporation process, with the goal of controlling the product liquor density. The proposed operational optimization framework dynamically finds the optimal control set points through the collaboration of an initial optimizer, a data-driven key process index predictor and a set point compensator. The process index predictor is developed based on the hierarchical Gaussian mixture model, which approximates the distribution of noisy and multimode process data. Then, the same model is used for finding the optimal compensation set points online, where an efficient sampling based stochastic search algorithm is proposed. The proposed method is tested using simulation and in a real industrial evaporation process showing superior performance. • A solution to the operation optimization of industrial evaporation process. • A hierarchal Gaussian mixture model (HGMM) for the prediction of key process index. • An adaptive top-down training algorithm of HGMM. • A sampling based optimization algorithm to find optimal set-points. [ABSTRACT FROM AUTHOR]

Voice conversion (VC) consists of digitally altering the voice of an individual to manipulate part of its content, primarily its identity, while maintaining the rest unchanged. Research in neural VC has accomplished considerable breakthroughs with the capacity to falsify a voice identity using a small amount of data with a highly realistic rendering. This paper goes beyond voice identity manipulation and presents an original neural architecture that allows the manipulation of voice attributes (e.g., gender and age). The proposed architecture is inspired by the fader network, transferring the same ideas to voice manipulation. The information conveyed by the speech signal is disentangled into interpretative voice attributes by means of minimizing adversarial loss to make the encoded information mutually independent while preserving the capacity to generate a speech signal from the disentangled codes. During inference for voice conversion, the disentangled voice attributes can be manipulated and the speech signal can be generated accordingly. For experimental evaluation, the proposed method is applied to the task of voice gender conversion using the freely available VCTK dataset. Quantitative measurements of mutual information between the variables of speaker identity and speaker gender show that the proposed architecture can learn gender-independent representation of speakers. Additional measurements of speaker recognition indicate that speaker identity can be recognized accurately from the gender-independent representation. Finally, a subjective experiment conducted on the task of voice gender manipulation shows that the proposed architecture can convert voice gender with very high efficiency and good naturalness. [ABSTRACT FROM AUTHOR]

Question matching is the fundamental task in retrieval-based dialogue systems which assesses the similarity between Query and Question. Unfortunately, existing methods focus on improving the accuracy of text similarity in the general domain, without adaptation to the financial domain. Financial question matching has two critical issues: (1) How to accurately model the contextual representation of a financial sentence? (2) How to accurately represent financial key phrases in an utterance? To address these issues, this paper proposes a novel Financial Knowledge Enhanced Network (FinKENet) that significantly injects financial knowledge into contextual text. Specifically, we propose a multi-level encoder to extract both sentence-level features and financial phrase-level features, which can more accurately represent sentences and financial phrases. Furthermore, we propose a financial co-attention adapter to combine sentence features and financial keyword features. Finally, we design a multi-level similarity decoder to calculate the similarity between queries and questions. In addition, a cross-entropy-based loss function is presented for model optimization. Experimental results demonstrate the effectiveness of the proposed method on the Ant Financial question matching dataset. In particular, the Recall score improves from 73.21% to 74.90% (1.69% absolute).

Road extraction from remote sensing imagery is a fundamental task in the field of image semantic segmentation. For this goal, numerous supervised deep learning techniques have been created, along with the employment of various loss functions that play a crucial role in determining the performances of supervised learning models. However, there is a lack of comprehensive analysis of the performance differences between the loss functions for road segmentation in remote sensing imagery. Therefore, this study conducts a comparative study of 12 well-known loss functions used widely in the field of image segmentation by training and evaluating the representative D-LinkNet network for road segmentation tasks with two publicly available remote sensing road datasets consisting of very high-resolution aerial and satellite images. The results show that different loss functions could lead to very different outcomes using the D-LinkNet, with varying focuses such as on overall model performances, precision, or recall. By dividing the loss functions into the distribution-based, region-based, and compound ones, we found that the region-based loss function type led to generally better model performances than the distribution-based one in terms of F1, IoU, and the road segmentation maps, with the compound loss function type being comparable to the region-based one. This paper eventually tries to offer suggestions for choosing the loss function that best suits the purposes of road segmentation-related studies.

Compared to cross-entropy in deep learning, triplet loss is less affected by the biased label information and widely used in fine-grained visual tasks. Especially in person re-identification (re-ID), triplet loss is improved with batch-hard sampling which only selects the hardest samples during the training process to reduce invalid triplets involved in the loss computation. The hardest samples' loss computation can provide a more intense gradient descent than raw samples. However, the batch-hard triplet loss discards multiple samples with important information, which can negatively impact feature learning. Besides, the hardest samples cause loss stuck problems frequently in training. In this work, we propose a balanced triplet loss for comprehensive feature learning and stable model convergence. The balanced triplet loss only mines the hardest negative samples of each category within a mini-batch. Compared with batch-hard triplet loss, it preserves the features of all the negative categories rather than one negative category with the hardest negative sample. It achieves a balance between triplet selection and information loss. The experiments show that our method can produce competitive results in re-ID tasks. In addition, we analyze the correlation between the intensity of data mining and the granularity of feature learning and further adapt the balanced triplet loss to general fine-grained image classification. The experiments prove the adapted balanced triplet loss also outperforms cross-entropy in multiple datasets of different scales. [ABSTRACT FROM AUTHOR]

When we compare the influences of two causes on an outcome, if the conclusion from every group is against that from the conflation, we think there is Simpson's Paradox. The Existing Causal Inference Theory (ECIT) can make the overall conclusion consistent with the grouping conclusion by removing the confounder's influence to eliminate the paradox. The ECIT uses relative risk difference Pd = max(0, (R − 1)/R) (R denotes the risk ratio) as the probability of causation. In contrast, Philosopher Fitelson uses confirmation measure D (posterior probability minus prior probability) to measure the strength of causation. Fitelson concludes that from the perspective of Bayesian confirmation, we should directly accept the overall conclusion without considering the paradox. The author proposed a Bayesian confirmation measure b* similar to Pd before. To overcome the contradiction between the ECIT and Bayesian confirmation, the author uses the semantic information method with the minimum cross-entropy criterion to deduce causal confirmation measure Cc = (R − 1)/max(R, 1). Cc is like Pd but has normalizing property (between −1 and 1) and cause symmetry. It especially fits cases where a cause restrains an outcome, such as the COVID-19 vaccine controlling the infection. Some examples (about kidney stone treatments and COVID-19) reveal that Pd and Cc are more reasonable than D; Cc is more useful than Pd. [ABSTRACT FROM AUTHOR]

Mining high-utility itemsets (HUIs) is one of the most important research topics in data mining because HUIs consider non-binary frequency values of items in transactions and different profit values for each item. However, setting appropriate minimum utility thresholds by trial and error is a tedious process for users. Thus, mining the top-k HUIs without setting a utility threshold is becoming an alternative to determine all the HUIs. In this paper, we propose two algorithms, called the top-k high-utility itemset mining based on cross-entropy method (TKU-CE) and TKU-CE+, for mining the top-k HUIs heuristically. The TKU-CE algorithm is based on cross-entropy, and implements top-k HUI mining using combinatorial optimization. The main idea of TKU-CE is to generate the top-k HUIs by gradually updating the probabilities of itemsets with high-utility values. TKU-CE+ optimizes TKU-CE in three respects. First, unpromising items are filtered by critical utility value, to reduce the computational burden in the initial stage. Second, a sample refinement strategy is used in each iteration, to reduce the computational burden in the iterative stage. Finally, smoothing mutation is proposed, to randomly generate some new itemsets in addition to those from previous iterations. Consequently, diversity of samples is improved, so that more actual top-k HUIs can be discovered with fewer iterations. Compared with state-of-the-art algorithms, TKU-CE and TKU-CE+ are easy to implement and avoid the computational costs that would be incurred by additional data structures and threshold-raising strategies. Extensive experimental results show that both algorithms are efficient, memory-saving, scalable, and can discover the most actual top-k HUIs. [ABSTRACT FROM AUTHOR]

In this article, we discuss an important concept of the generalized Radon–Nikodym derivatives using Choquet integral and its correct treatment. It is consequently used in the definition of generalized ϕ -divergence for fuzzy measures with the emphasis on its proper use. We give several results concerning basic properties of ϕ -divergence. In both concepts, we focus mainly on sufficient conditions for which standard properties copy the additive case. As it is shown, the key role in most of the relevant properties for both derivative as well as divergence is the subadditivity of fuzzy measure. [ABSTRACT FROM AUTHOR]

The study of leukemia classification using deep learning techniques has been conducted by multiple research teams worldwide. Although deep convolutional neural networks achieved high quality of sick vs. healthy patient discrimination, their inherent lack of human interpretability of the decision-making process hinders the adoption of deep learning techniques in medicine. Research involving deep learning proved that distinguishing between healthy and sick patients using microscopic images of lymphocytes is possible. However, it could not provide information on the intermediate steps in the diagnosis process. As a result, despite numerous examinations, it is still unclear whether the lymphocyte is the only object in the microscopic picture containing leukemia-related information or if the leukocyte's surroundings also contain the desired information. In this work, entropy measures and machine learning models were applied to study the informativeness of both whole images and lymphocytes' surroundings alone for Leukemia classification. This work aims to provide human-interpretable features marking the probability of sickness occurrence. The research stated that the hue distribution of images with lymphocytes obfuscated alone is informative enough to facilitate 93.0% accuracy in healthy vs. sick classification. The research was conducted on the ALL-IDB2 dataset. [ABSTRACT FROM AUTHOR]

The Dual Credit Policy is an important policy to promote the development of new energy vehicles unique to China. There is a lack of research that intuitively reflects the impact of the Dual Credit Policy on industrial development through an industry-based factual comparison of this policy. Based on the Taylor expansion and Cross-Entropy description, this article obtains the development regression function by the quantitative analysis of five indicators—the number of new energy vehicle-related patents, sales volume, production volume, the number of newly registered enterprises, infrastructure construction (the number of charging piles) before and after the implementation of the policy, and describes them quantitatively using the Taylor expansion to obtain the CPTI index. The CPCEI index is obtained by calculating the Cross-Entropy of the distribution of each indicator before and after policy implementation. The above two indices were compared for the growth trend and growth quantity, respectively. Finally, the following conclusions were obtained: 1. the Dual Credit Policy is more significantly promoted at the market level than the impact on the technical level; 2. although there is also incentive in infrastructure construction, it cannot fully react to the market demand; 3. the number of start-up’s operating in the new energy field increases, but the overall growth trend gradually slows down and fails to significantly change the existing structure of the market. This study suggests that the government should launch a special incentive policy for charging piles, and new energy manufacturers should expand their production capacity to meet the market demand.