Asymmetric Loss Research Articles

Weighted Asymmetric Loss for Multi-Label Text Classification on Imbalanced Data

Weighted Asymmetric Loss for Multi-Label Text Classification on Imbalanced Data

Analysis of a new jointly hybrid censored Rayleigh populations

<abstract><p>When a researcher wants to perform a life-test comparison study of items made by two separate lines inside the same institution, joint censoring strategies are particularly important. In this paper, we present a new joint Type-Ⅰ hybrid censoring that enables an experimenter to stop the investigation as soon as a pre-specified number of failures or time is first achieved. In the context of newly censored data, the estimates of the unknown mean lifetimes of two different Rayleigh populations are acquired using maximum likelihood and Bayesian inferential techniques. The normality characteristic of classical estimators is used to offer asymptotic confidence interval bounds for each unknown parameter. Against gamma conjugate priors, the Bayes estimators and related credible intervals are gathered about symmetric and asymmetric loss functions. Since classical and Bayes estimators are acquired in closed form, simulation tests can be easily made to evaluate the effectiveness of the proposed methodologies. The efficiency of the suggested approaches is examined in terms of four metrics, namely: Root mean squared error, average relative absolute bias, average confidence length, and coverage probability. To demonstrate the applicability of the offered approaches to real events, two real applications employing data sets from the engineering area are analyzed. As a result, when the experimenter's primary goal is to complete the test as soon as the total number of failures or the threshold period is recorded, the numerical results reveal that the recommended strategy is adaptable and very helpful in completing the study.</p></abstract>

Reliability analysis for two populations Nadarajah-Haghighi distribution under Joint progressive type-II censoring

<abstract><p>In order to evaluate the competitive advantages and dependability of two products in a competitive environment, comparative lifespan testing becomes essential. We examine the inference problems that occur when two product lines follow the Nadarajah-Haghighighi distribution in the setting of joint type-II censoring. In the present study, we derived the maximum likelihood estimates for the Nadarajah-Haghighi population parameters. Additionally, a Fisher information matrix was constructed based on these maximum likelihood estimations. Furthermore, Bayesian estimators and their corresponding posterior risks were calculated, considering both gamma and non-informative priors under symmetric and asymmetric loss functions. To assess the performance of the overall parameter estimators, we conducted a Monte Carlo simulation using numerical methods. Lastly, a real data analysis was carried out to validate the accuracy of the models and methods discussed.</p></abstract>

ALPACA: An Asymmetric Loss Prediction Algorithm for Channel Adaptation Based on a Convolutional-Recurrent Neural Network in URLLC Systems

ALPACA: An Asymmetric Loss Prediction Algorithm for Channel Adaptation Based on a Convolutional-Recurrent Neural Network in URLLC Systems

A Case Report of Leprosy Presenting as Multiple Mononeuropathy with Hypoesthesia

Leprosy is a chronic infectious disease caused by Mycobacterium leprae. It can lead to damage of the nerve. Although the incidence of leprosy is very low in South Korea, a large number of people are immigrating to South Korea from countries with a high prevalence of leprosy. We report a case of leprosy confirmed by nerve biopsy. The patient was from Nepal who presented with progressive and asymmetric sensory loss. Leprosy can be considered as a differential diagnosis in patients with progressive and asymmetric sensory loss, especially when patients are from leprosy endemic countries.

Estimation of multicomponent stress–strength reliability for exponentiated Gumbel distribution

In this paper, the stress–strength reliability R s , k of a multicomponent s-out-of-k system for exponentiated Gumbel distribution is considered. An s-out-of-k system means a system with total k components and the system can survive only when atleast s of the total components function properly. The ability of the system to overcome the experiencing stress with its strength is termed as its stress–strengh reliability. The maximum likelihood estimator and Bayes estimator for R s , k are obtained. The Bayes estimators are obtained using Markov chain Monte Carlo(MCMC) method under both symmetric and asymmetric loss functions. The loss functions we considered are squared error loss function, LINEX loss function and entropy loss function. The asymptotic, bootstrap and highest posterior density(HPD) confidence intervals for R s , k are also obtained. A simulation study is conducted for evaluating the efficiency of the estimators derived in this paper. Real data sets are also considered for illustration.

Bayesian and non-Bayesian analysis for the lifetime performance index based on generalized order statistics from Pareto distribution

<p>Modern businesses depend on efficient management and evaluation of product quality performance to assure that they are on the right track, and process capability analysis is used to gauge business performance in practice. Consequently, the lifetime performance index (LPI) , where is the lower specification limit, is used to gauge a process potential and performance. This paper examines distinct estimators of under Pareto distribution using generalized order statistics (GOS), which is very helpful in a variety of real-world applications. Results for progressive type II censoring (PTIIC) and first-failure censoring are two particular situations. Using symmetric and asymmetric loss functions, the Bayesian estimator was built, then utilized to produce the hypothesis testing technique. A simulation study and real data analysis have been investigated to study the behavior of different estimates for under different schemes, namely PTIIC and the progressive first failure censored scheme.</p>

Assessing systemic risk and connectedness among dirty and clean energy markets from the quantile and expectile perspectives

In response to environmental and climate change issues in recent decades, clean energy resources have been promoted as substitutes for fossil fuel-based dirty energy resources. While receiving great attention from environmentally conscious investors and policy-makers, the markets for the former energies could be affected by financial instability in the markets for the latter, which could lead to systemic risk and directional connectedness. In this paper, we aimed to assess potential systemic risk in such markets by employing conditional quantile- and conditional expectile-based Value-at-Risk (henceforth CoVaR and CoEVaR, respectively). Due to their definition through expected conditional asymmetric loss functions, we adopted Kabaila’s efficient method for the computation of conditional expectations and proposed a backtesting technique based on Diebold and Mariano’s test. Using Delta-CoVaR and Delta-CoEVaR, we constructed conditional tail risk networks to describe the directional tail risk connectedness. We revealed that the CoVaR and CoEVaR for almost all pairs of energy markets had the best forecast accuracy when computed using a copula exhibiting symmetric tail dependence with asymmetric and heavy-tailed margins. Compared to CoVaR, CoEVaR possessed a relatively better conditional coverage performance in several cases, complementing its advantage of being more sensitive to (extreme) loss magnitude. In addition, we found crude oil, heating oil, and sectoral clean energy indices to be highly interconnected, but the connectedness weakened following the 2015 Paris Climate Agreement. In particular, crude oil and the renewable energy sector played a consistent role as the most systemically important net receiver and net transmitter, respectively. These results provided important implications for policy-makers and investors aware of systemic risk and connectedness among dirty and clean energy markets.

Benefits of unilateral cochlear implantation in adults with asymmetric hearing loss: Audiologic and patient-related outcome measures

PurposeTo investigate the benefits of cochlear implantation in adults with single-sided deafness (SSD) and asymmetric hearing loss (AHL). Study designProspective within-subjects repeated-measures. SettingTwo tertiary cochlear implant centers. PatientsFourteen adults with severe-to-profound sensorineural hearing loss in the worse hearing ear and up to moderate SNHL in the better hearing ear. InterventionCochlear implantation in the worse hearing ear. Main outcome measuresConsonant-nucleus-consonant (CNC) test, AzBio sentence test in noise, and lateralization testing were conducted preoperatively and at 3-, 6-, and 12-months post-activation. Patient-related outcomes were measured using the Speech, Spatial, and Qualities of Hearing Scale and Glasgow Benefit Inventory. Tinnitus Handicap Inventory was administered to subjects with tinnitus. ResultsMean length of hearing loss in the worse hearing ear was 3.5 years. The mean CNC change scores from baseline were 54.8, 55.9, and 58.9 percentage points at 3-, 6-, and 12-months (p < 0.001). AzBio sentence test in noise demonstrated improved scores in all spatial configurations, although statistically significant in S0N0 (speech front, noise front) only. Lateralization testing showed significant improvement of 22.9, 24.5, and 24.0 percentage points at 3-, 6-, and 12 months post-activation (p = 0.002). All patient-related outcome measures revealed significant improvement. ConclusionThis study demonstrates improved speech perception in noise, sound lateralization, quality of life, and reduction in tinnitus perception in adults with SSD/AHL who undergo cochlear implantation. Our results add to the growing body of evidence that cochlear implant should be offered to this population.

Auditory capacity of the better-hearing ear in asymmetric hearing loss

PurposeOur aim was to investigate the course of the hearing capacity of the better-hearing ear in single-sided deafness (SSD) and asymmetric hearing loss (AHL) over time, in a multicenter study.MethodsWe included 2086 pure-tone audiograms from 323 patients with SSD and AHL from four hospitals and 156 private practice otorhinolaryngologists. We collected: age, gender, etiology, duration of deafness, treatment with CI, number and monosyllabic speech recognition, numerical rating scale (NRS) of tinnitus intensity, and the tinnitus questionnaire according to Goebel and Hiller. We compared the pure tone audiogram of the better-hearing ear in patients with SSD with age- and gender-controlled hearing thresholds from ISO 7029:2017.ResultsFirst, individuals with SSD showed a significantly higher hearing threshold from 0.125 to 8 kHz in the better-hearing ear compared to the ISO 7029:2017. The duration of deafness of the poorer-hearing ear showed no relationship with the hearing threshold of the better-hearing ear. The hearing threshold was significantly higher in typically bilaterally presenting etiologies (chronic otitis media, otosclerosis, and congenital hearing loss), except for Menière’s disease. Second, subjects that developed AHL did so in 5.19 ± 5.91 years and showed significant reduction in monosyllabic word and number recognition.ConclusionsIndividuals with SSD show significantly poorer hearing in the better-hearing ear than individuals with NH from the ISO 7029:2017. In clinical practice, we should, therefore, inform our SSD patients that their disease is accompanied by a reduced hearing capacity on the contralateral side, especially in certain etiologies.

Multi-view fusion-based local-global dynamic pyramid convolutional cross-tansformer network for density classification in mammography

Object. Breast density is an important indicator of breast cancer risk. However, existing methods for breast density classification do not fully utilise the multi-view information produced by mammography and thus have limited classification accuracy. Method. In this paper, we propose a multi-view fusion network, denoted local-global dynamic pyramidal-convolution transformer network (LG-DPTNet), for breast density classification in mammography. First, for single-view feature extraction, we develop a dynamic pyramid convolutional network to enable the network to adaptively learn global and local features. Second, we address the problem exhibited by traditional multi-view fusion methods, this is based on a cross-transformer that integrates fine-grained information and global contextual information from different views and thereby provides accurate predictions for the network. Finally, we use an asymmetric focal loss function instead of traditional cross-entropy loss during network training to solve the problem of class imbalance in public datasets, thereby further improving the performance of the model. Results. We evaluated the effectiveness of our method on two publicly available mammography datasets, CBIS-DDSM and INbreast, and achieved areas under the curve (AUC) of 96.73% and 91.12%, respectively. Conclusion. Our experiments demonstrated that the devised fusion model can more effectively utilise the information contained in multiple views than existing models and exhibits classification performance that is superior to that of baseline and state-of-the-art methods.

Business applications and state‐level stock market realized volatility: A forecasting experiment

AbstractWe analyze the predictive value of (the surprise component of) state‐level business applications, as a proxy of local investor sentiment, for the state‐level realized US stock‐market volatility. We use high‐frequency data for the period from September 2011 to October 2021 to compute realized volatility. Using an extended version of the popular heterogeneous autoregressive realized volatility model and accounting for the possibility that users of forecasts have an asymmetric loss function, we show that business applications tend to have predictive value for realized state‐level stock‐market volatility, as well as for upside (“good”) and downside (“bad”) realized volatility, for users of forecasts who suffer a larger loss from an underprediction of realized volatility than from an overprediction of the same (absolute) seize, after controlling for realized moments (realized skewness, realized kurtosis, realized jumps, and realized tail risks). We also highlight that the COVID‐19 period is a major driver of our empirical results.

Bayesian and E-Bayesian estimation based on constant-stress partially accelerated life testing for inverted Topp–Leone distribution

Abstract Accelerated or partially accelerated life tests are particularly significant in life testing experiments since they save time and cost. Partially accelerated life tests are carried out when the data from accelerated life testing cannot be extrapolated to usual conditions. The constant-stress partially accelerated life test is proposed in this study based on a Type-II censoring scheme and supposing that the lifetimes of units at usual conditions follow the inverted Topp–Leone distribution. The Bayes and E-Bayes estimators of the distribution parameter and the acceleration factor are derived. The balanced squared error loss function, which is a symmetric loss function, and the balanced linear exponential loss function, which is an asymmetric loss function, are considered for obtaining the Bayes and E-Bayes estimators. Based on informative gamma priors and uniform hyper-prior distributions, the estimators are obtained. Finally, the performance of the proposed Bayes and E-Bayes estimates is evaluated through a simulation study and an application using real datasets.

Reliability and the engineering applications of the generalized half-normal model via an adaptive progressive hybrid censored mechanism

Recently, the generalized half-normal distribution with decreasing, increasing, and bathtub hazard function shapes was proposed, making it a more applicable, reliable, and flexible lifespan model. The task of estimating the unknown parameters and reliability features of the generalized half-normal distribution is looked at using adaptive progressively type-II hybrid censored data. The maximum likelihood and Bayesian estimation methods are both considered for this purpose. Two approximated confidence intervals, Bayes and highest posterior density intervals, are acquired for the various parameters. The Bayes estimates are obtained based on symmetric and asymmetric loss functions under the assumption of independent gamma priors. The Markov chain Monte Carlo approach is used to compute Bayes estimates as well as the various Bayes intervals. Monte Carlo experiments are used for assessing the efficiency of the various approaches. Finally, analysis is performed on two actual-life engineering datasets.

Deep Learning for Precise Robot Position Prediction in Logistics

This study presents an interdisciplinary investigation at the nexus of mechanical engineering and computer science, aimed at advancing the field of logistics automation. In response to the escalating demands of global cargo transportation, the integration of these disciplines assumes paramount importance. Conducted within the domain of Dortmund University of Technology’s Material Flow and Warehousing Chair, this research focuses on the precise control of robots, a task contingent on accurate positional information. Leveraging a controlled internal logistics precinct, the study delves into the transformation of raw sensor data, comprising accelerometers, gyroscopes, and magnetometers, into precise position predictions. This process entails meticulous data preprocessing, encompassing synchronization and calibration procedures, yielding crucial parameters such as absolute velocity and accelerations along both parallel and perpendicular axes. The study employs deep learning, specifically a 2D Convolutional Neural Network (2D-CNN), for predictive modeling. This architecture excels in extracting intricate spatial features from sensor data. Training is conducted under the guidance of an Asymmetric Gaussian loss function, custom-tailored to accommodate the idiosyn- crasies of real-world sensor data. The results evince the efficacy of this approach, evidenced by remarkably low mean squared errors in predicting robot positions. Beyond its immediate applications in logistics automation, this research underscores the potential of interdisciplinary collaboration in addressing complex sensor data challenges.

Symptom: Asymmetric Mixed Hearing Loss

Symptom: Asymmetric Mixed Hearing Loss

Detecting Persuasion Attempts on Social Networks: Unearthing the Potential of Loss Functions and Text Pre-Processing in Imbalanced Data Settings

The rise of social networks and the increasing amount of time people spend on them have created a perfect place for the dissemination of false narratives, propaganda, and manipulated content. In order to prevent the spread of disinformation, content moderation is needed. However, manual moderation is unfeasible due to the large amount of daily posts. This paper studies the impact of using different loss functions on a multi-label classification problem with an imbalanced dataset, consisting of 20 persuasion techniques and only 950 samples, provided by SemEval’s 2021 Task 6. We used machine learning models, such as Naive Bayes and Decision Trees, and a custom deep learning architecture, based on DistilBERT and Convolutional Layers. Overall, the machine learning models achieved far worse results than the deep learning model, using Binary Cross Entropy, which we considered our baseline deep learning model. To address the class imbalance problem, we trained our model using different loss functions, such as Focal Loss and Asymmetric Loss. The latter providing the best results, particularly for the least represented classes.

An economic-statistical design of synthetic Tukey’s control chart with Taguchi’s asymmetric loss functions under log-normal distribution

The synthetic Tukey’s control chart (denoted by Syn-TCC) applies a single observation in the process monitoring and is relatively robust to outliers. For a complex process, the probability distribution of quality characteristic must be carefully identified and the loss of quality characteristic, which is defined as the deviation from the target value, must be accurately estimated so that the quality state of the process may be more comprehensively described. However, the literature rarely studied the economic design of the Syn-TCC with Taguchi’s loss function under non normality. In the present study, it is assumed that the quality characteristic follows a log-normal distribution and Taguchi’s asymmetric linear and quadratic loss functions are involved to develop the economic-statistical design of the Syn-TCC for multiple assignable causes. A case study is presented to illustrate the application of this economic-statistical design of the Syn-TCC to an IC package industry. The sensitivity analysis of this case study reveals that a larger production lot size and larger two coefficients of loss functions generally lead to a higher total expected cost.

Mineral identification based on natural feature-oriented image processing and multi-label image classification

Artificial intelligence (AI) technology has significant potential in Earth sciences, particularly in mineral identification for industrial exploration, geological mapping, and archaeological research. However, traditional methods are time-consuming, expensive, and complex. And existing mineral identification methods based on mineral photos face several critical challenges, including lack of consideration for natural image features captured in real environments, limitations of single-label classification which does not align with multi-mineral occurrences in nature, and growing computational complexity as the number of identifiable mineral labels increases. Therefore, this paper proposes an efficient mineral identification model based on multi-label image classification, focusing on natural environmental features. First, realistic feature datasets are created by simulating mineral photos in real environments. Then, the model uses the query-label (Query2Label) framework, with MaxViT-T (Multi-Axis Vision Transformer-Tiny) as the feature extraction network and the asymmetric loss function. Knowledge distillation is employed to improve identification accuracy while reducing computational complexity. The proposed model achieves an impressive average identification accuracy of 84.74% on a dataset of 495,756 mineral photos, surpassing existing models like ResNet-101, ML-GCN (Multi-Label Graph Convolutional Network), and SRN (Spatial Regularization Net). It maintains a lower parameter count and computational complexity. In the end, ablation experiments demonstrate the effectiveness of each optimization scheme.

Evaluating Autoencoders as a Dimensionality Reduction Mechanism to Support Clustering Brazilian Agricultural Diversity

Brazilian agricultural production presents high spatial diversity, challenging the conception of public policies. This article proposes an approach for grouping Brazilian municipalities according to their agricultural production. We combine a feature extraction using autoencoders and clustering based on k-means and Self-Organizing Maps. We used panel data from IBGE’s annual estimates of the production value of permanent and temporary crops, animal products, aquaculture, plant extractivism, forestry, planted areas, and herd population between 1999 and 2018. We analyzed different structures of simple stacked and incomplete autoencoders, varying the number of layers and neurons in each, and evaluated the asymmetric exponential linear loss function to handle the sparse data. We applied the Isomap, Kernel PCA, Truncated SVD, and MDS dimensionality reduction methods for comparative purposes. Results showed that the autoencoders could extract characteristics from the transformed raw data to allow the clustering of municipalities to reveal regional and evenintra-regional patterns. The autoencoders improved comparative performance as the intrinsic dimensionality increased.