A Robust Framework for Multimodal Sentiment Analysis with Noisy Labels Generated from Distributed Data Annotation

Multimodal sentiment analysis utilizes multimodal data such as text,facial expressions and voice to detect people’s attitudes.With the advent of distributed data collection and annotation,we can easily obtain and share such multimodal data.However,due to professional discrepancies among annotators and lax quality control,noisy labels might be introduced.Recent research suggests that deep neural networks(DNNs)will overfit noisy labels,leading to the poor performance of the DNNs.To address this challenging problem,we present a Multimodal Robust Meta Learning framework(MRML)for multimodal sentiment analysis to resist noisy labels and correlate distinct modalities simultaneously.Specifically,we propose a two-layer fusion net to deeply fuse different modalities and improve the quality of the multimodal data features for label correction and network training.Besides,a multiple meta-learner(label corrector)strategy is proposed to enhance the label correction approach and prevent models from overfitting to noisy labels.We conducted experiments on three popular multimodal datasets to verify the superiority of ourmethod by comparing it with four baselines.

Radio Frequency Fingerprinting Identification Using Semi-Supervised Learning with Meta Labels

Radio frequency fingerprinting(RFF)is a remarkable lightweight authentication scheme to support rapid and scalable identification in the internet of things(IoT)systems.Deep learning(DL)is a critical enabler of RFF identification by leveraging the hardware-level features.However,traditional supervised learning methods require huge labeled training samples.Therefore,how to establish a highperformance supervised learning model with few labels under practical application is still challenging.To address this issue,we in this paper propose a novel RFF semi-supervised learning(RFFSSL)model which can obtain a better performance with few meta labels.Specifically,the proposed RFFSSL model is constituted by a teacher-student network,in which the student network learns from the pseudo label predicted by the teacher.Then,the output of the student model will be exploited to improve the performance of teacher among the labeled data.Furthermore,a comprehensive evaluation on the accuracy is conducted.We derive about 50 GB real long-term evolution(LTE)mobile phone’s raw signal datasets,which is used to evaluate various models.Experimental results demonstrate that the proposed RFFSSL scheme can achieve up to 97%experimental testing accuracy over a noisy environment only with 10%labeled samples when training samples equal to 2700.

Factors Related to the Reading Nutritional Labels by Consumers

This study aimed to evaluate the features related to consumers’ reading nutritional labels in a city in the interior of the São Paulo State, Brazil. A questionnaire was answered by 100 consumers of a supermarket chain, sociodemographic information and data related to label reading habits were collected. Tables with percentage values and bar graphs were used. Chi-square tests and logistic regression models were performed to verify the association between the variables and the label reading habits. The factors that showed significant associations with the reading labels were gender, ease to understand the labels and access to their information (p 0.10). People who had already read labels reported to have more difficulty to understand the information contained on them, and people who had already received instructions on the labels were three and a half times more likely to read the instructions contained on them than those who hadn’t received any guidance. This study points to the need to expand the disclosure to consumers about the contents present on the labels, through more accessible language, so that the labels fulfill their role to instruct consumers in their choices.

Research on Key Technologies of Electronic Shelf Labels Based on LoRa

The demand for Electronic Shelf Labels(ESL),according to the Internet of Things(IoT)paradigm,is expected to grow considerably in the immediate future.Various wireless communication standards are currently contending to gain an edge over the competition and provide the massive connectivity that will be required by a world in which everyday objects are expected to communicate with each other.Low-Power Wide-Area Networks(LPWANs)are continuously gaining momentum among these standards,mainly thanks to their ability to provide long-range coverage to devices,exploiting license-free frequency bands.The main theme of this work is one of the most prominent LPWAN technologies,LoRa.The purpose of this research is to provide long-range,less intermediate node,less energy dissipation,and a cheaper ESL system.Much research has already been done on designing the LoRaWAN network,not capable to make a reliable network.LoRa is using different gateways to transmit the same data,collision,data jamming,and data repetition are expected.According to the transmission behavior of LoRa,50%of data is lost.In this paper,the Improved Backoff Algorithm with synchronization technique is used to decrease overlapping and data loss.Besides,the improved Adaptive Data Rate algorithm(ADR)avoids the collision in concurrently transmitted data by using different Spreading Factors(SFs).The allocation of SF has the main role in designing LoRa based network to minimize the impact of the intra-interference,cost function,and Euclidean distance.For this purpose,the K-means machine learning algorithm is used for clustering.The data rate model is using an intra-slicing technique based on Maximum Likelihood Estimation(MLE).The data rate model includes three critical communication slices,High Critical Communication(HCC),Medium Critical Communication(MCC),and Low Critical Communication(LCC),having the specified number of End devices(EDs),payload budget delay,and data rate.Finally,different combinations of gateways are used to build ESL for 200 electronic shelf labels.

From indie to major labels, denim abounds as birthday of the blue jean nears

The year 1873 was a busy one for San Francisco. That was the year the University of California opened its first medical school in the City by the Bay. San Francisco’s cable cars first began running. And the blue jean was born after tailor Jacob Davis and fabric supplier Levi Strauss received the patent for their copper-riveted denim cotton bottoms. Now, the UCSF School of Medicine is one of the top-ranked in the country. The cable cars are an iconic form of transit in the city. And the blue jean, despite generations of trends and changes in taste, remains a powerhouse in the apparel industry, an item that’s worn as often by kids and fashion models as soccer dads and rock stars.

Labels X1．6用颜色区分你的文件夹

使用过Mac OS9的读者应该记得9系统是可以用颜色来区分文件夹的。但当我们将系统升级到Mac OS X后，我们只可以将文件名增加背景颜色。不过近日小编发现了一个叫Labels X的小工具，当你使用它后，你会发现将文件夹图标整个加上颜色，会更容易区分它们。采用这种方法后，即使屏幕上的文件再多，你也可以对不同种类的文件目录了如指掌。

杜邦TM赛丽FAST系统助力Hub Labels从优质到卓越的蜕变

座落于美国马里兰州黑格斯敦的Hub Labels成立于1978年,作为首批引入四色印刷、计算机直接制版、数码印刷及无齿轮印刷的标签公司之一,Hub Labels可为客户提供各式各样的定制标签印刷,主要有传统压敏标签、无底纸标签、直邮产品三大类产品。

Orthogonal Stacked Spectral Coding Labels for Fast Packets Routing over Optical MPLS Network

Multi-protocol label switching（MPLS） has the advantage of high efficiency in the second layer, which improves the performance of data packets routing. In this paper, a new structure to implement optical MPLS is proposed. We construct a code family for spectral-amplitude coding（SAC） labels in the optical MPLS networks. SAC labels are suitable for optical packet switching because they can be constructed and recognized quickly at each router. We use the label stacking to provide hierarchical routing to avoid swapping labels at each forwarding node and reduce system complexity. However, the phase-induced intensity noise（PIIN） appears due to the incoherent property of the light source when the stacked labels set makes the correlation decoding with the local node label,which degrades system performance.

Optical Packet Switching Scheme Using Multi-optical-code Labels

Proposed is a novel optical code(OC) label switching scheme in which an optical label is constructed by multiple parallel optical codes.The performances of splitting loss and BER are simulated and analyzed.Simulation results show that the proposed label can be correctly recognized to perform packet switching.Compared with reported schemes using one OC as a label,the splitting loss in our proposal is lowered.

Quality Assessment of Training Data with Uncertain Labels for Classification of Subjective Domains

In order to improve the performance of classifiers in subjective domains, this paper defines a metric to measure the quality of the subjectively labelled training data (QoSTD) by means of K-means clustering. Then, the QoSTD is used as a weight of the predicted class scores to adjust the likelihoods of instances. Moreover, two measurements are defined to assess the performance of the classifiers trained by the subjective labelled data. The binary classifiers of Traditional Chinese Medicine (TCM) Zhengs are trained and retrained by the real-world data set, utilizing the support vector machine (SVM) and the discrimination analysis (DA) models, so as to verify the effectiveness of the proposed method. The experimental results show that the consistency of likelihoods of instances with the corresponding observations is increased notable for the classes, especially in the cases with the relatively low QoSTD training data set. The experimental results also indicate the solution how to eliminate the miss-labelled instances from the training data set to re-train the classifiers in the subjective domains.

Learning multi-tasks with inconsistent labels by using auxiliary big task

Multi-task learning is to improve the performance of the model by transferring and exploiting common knowledge among tasks.Existing MTL works mainly focus on the scenario where label sets among multiple tasks(MTs)are usually the same,thus they can be utilized for learning across the tasks.However,the real world has more general scenarios in which each task has only a small number of training samples and their label sets are just partially overlapped or even not.Learning such MTs is more challenging because of less correlation information available among these tasks.For this,we propose a framework to learn these tasks by jointly leveraging both abundant information from a learnt auxiliary big task with sufficiently many classes to cover those of all these tasks and the information shared among those partiallyoverlapped tasks.In our implementation of using the same neural network architecture of the learnt auxiliary task to learn individual tasks,the key idea is to utilize available label information to adaptively prune the hidden layer neurons of the auxiliary network to construct corresponding network for each task,while accompanying a joint learning across individual tasks.Extensive experimental results demonstrate that our proposed method is significantly competitive compared to state-of-the-art methods.

基于Yolov5的交通信号灯智能识别程序开发

交通信号检测是智能汽车识别交通环境的一项重要辅助技术,现有的算法能够解决单一交叉口环境下的信号检测问题,但需要在十字路口的复杂交通环境中提高算法的精度和干扰可靠性。文章以one-stage目标检测算法Yolov5的应用为研究基础,实现多场景下的交通信号灯自动检测与识别,使用Labeling进行图片标注,通过镜像、裁剪、反转、等运行增强数据集,不断地调参实验与迭代模型训练,目标检测精度达到80%。

近红外无创血糖浓度的Label Sensitivity算法和支持向量机回归

近红外光谱分析技术在生物医学工程领域具有广阔应用前景。无创且持续性地测量能实时监控人体血糖水平,给糖尿病患者带来极大便利性、提高生存质量、降低糖尿病并发症发生率具有很大的社会效益。无创血糖监测的想法提出较早,但仍然存在预测精度低、预测值与标签值相关性不高等难点,至今没有达到临床要求。近年来,光谱检测技术发展迅猛且机器学习技术在智能信息处理方面具有明显优势,两者结合可以有效提高人体无创血糖医学监测模型的精度和普适性。提出了一种标签敏感度算法(LS),并结合支持向量机方法建立了人体血糖含量预测模型。使用近红外光谱仪采集了4名志愿者食指处动态血液光谱数据(每名志愿者28组数据),并使用多元散射矫正(MSC)方法消除了部分光散射的影响。考虑血糖对不同波长光的吸收有差异,提出了基于血糖浓度标签差的特征波长挑选方法,并构建了标签敏感度支持向量机(LSSVR)预测模型。设计实验,对比该模型与偏最小二乘回归(PLSR)和区分度支持向量机(FSSVR)算法。结果表明,LS算法的最佳特征波长数为32,经特征波长选择后的LSSVR表现最佳,其均方误差降低至0.02 mmol·L^(-1),明显优于全谱段PLSR模型,血糖浓度的预测值与标签值的相关系数提升至99.8%,预测值全部位于可容许误差的克拉克网格A区内。LSSVR模型的优异表现为早日实现血糖的无创监测提供了新思路。

Consumer willingness to pay for eco-labels in China:A choice experiment approach

The use of eco-labels has become increasingly popular in China.This study aims to understand whether Chinese consumers are willing to pay for eco-labels and how their willingness to pay(WTP)is determined.Although we find that Chinese consumers are willing to pay more for products that are labeled as having greater energy efficiency(Energy Efficiency Label)or as having been produced using more environmentally-friendly production processes(Environmental Label),some practices have significantly impaired the effectiveness of these labels,e.g.the burgeoning use of eco-labels has led to label confusion;public trust towards eco-labels.The policy implications are discussed in this study.

Label Recovery and Trajectory Designable Network for Transfer Fault Diagnosis of Machines With Incorrect Annotation

The success of deep transfer learning in fault diagnosis is attributed to the collection of high-quality labeled data from the source domain.However,in engineering scenarios,achieving such high-quality label annotation is difficult and expensive.The incorrect label annotation produces two negative effects:1)the complex decision boundary of diagnosis models lowers the generalization performance on the target domain,and2)the distribution of target domain samples becomes misaligned with the false-labeled samples.To overcome these negative effects,this article proposes a solution called the label recovery and trajectory designable network(LRTDN).LRTDN consists of three parts.First,a residual network with dual classifiers is to learn features from cross-domain samples.Second,an annotation check module is constructed to generate a label anomaly indicator that could modify the abnormal labels of false-labeled samples in the source domain.With the training of relabeled samples,the complexity of diagnosis model is reduced via semi-supervised learning.Third,the adaptation trajectories are designed for sample distributions across domains.This ensures that the target domain samples are only adapted with the pure-labeled samples.The LRTDN is verified by two case studies,in which the diagnosis knowledge of bearings is transferred across different working conditions as well as different yet related machines.The results show that LRTDN offers a high diagnosis accuracy even in the presence of incorrect annotation.

Interactive visualization of 3D lunar model with texture and labels,using Chang'E-1 data

A lunar model with real texture can be obtained by mapping texture onto the lunar mesh,but the convergence in the polar regions of lunar model is a problem.In this paper,we build a 3D lunar model and solve this problem by texture partitioning and transforming.The whole lunar map is divided into four images and the polar images are transformed to circular textures before mapped to the semi-regular(SR) lunar mesh which is obtained through denoising,triangulating,subdividing and resampling the laser altimetry(LAM) data.Hundreds of lunar labels are classed into three levels and added gradually to the lunar model considering the distance between the viewpoint and the moon center.Through some techniques such as mip-map and view-dependent,the lunar model with textures and labels can be interactively browsed on a personal computer(PC) in real time.

Data fusing and joint training for learning with noisy labels

It is well known that deep learning depends on a large amount of clean data.Because of high annotation cost,various methods have been devoted to annotating the data automatically.However,a larger number of the noisy labels are generated in the datasets,which is a challenging problem.In this paper,we propose a new method for selecting training data accurately.Specifically,our approach fits a mixture model to the per-sample loss of the raw label and the predicted label,and the mixture model is utilized to dynamically divide the training set into a correctly labeled set,a correctly predicted set,and a wrong set.Then,a network is trained with these sets in the supervised learning manner.Due to the confirmation bias problem,we train the two networks alternately,and each network establishes the data division to teach the other network.When optimizing network parameters,the labels of the samples fuse respectively by the probabilities from the mixture model.Experiments on CIFAR-10,CIFAR-100 and Clothing1M demonstrate that this method is the same or superior to the state-of-the-art methods.

Hybrid Bayesian estimation tree learning with discrete and fuzzy labels

Classical decision tree model is one of the classical machine learning models for its simplicity and effectiveness in applications. However, compared to the DT model, probability estimation trees （PETs） give a better estimation on class probability. In order to get a good probability estimation, we usually need large trees which are not desirable with respect to model transparency. Linguistic decision tree （LDT） is a PET model based on label semantics. Fuzzy labels are used for building the tree and each branch is associated with a probability distribution over classes. If there is no overlap between neighboring fuzzy labels, these fuzzy labels then become discrete labels and a LDT with discrete labels becomes a special case of the PET model. In this paper, two hybrid models by combining the naive Bayes classifier and PETs are proposed in order to build a model with good performance without losing too much transparency. The first model uses naive Bayes estimation given a PET, and the second model uses a set of small-sized PETs as estimators by assuming the independence between these trees. Empirical studies on discrete and fuzzy labels show that the first model outperforms the PET model at shallow depth, and the second model is equivalent to the naive Bayes and PET.