Debiaser for Multiple Variables to enhance fairness in classification tasks

Nowadays assuring that search and recommendation systems are fair and do not apply discrimination among any kind of population has become of paramount importance. This is also highlighted by some of the sustainable development goals proposed by the United Nations. Those systems typically rely on machine learning algorithms that solve the classification task. Although the problem of fairness has been widely addressed in binary classification, unfortunately, the fairness of multi-class classification problem needs to be further investigated lacking well-established solutions. For the aforementioned reasons, in this paper, we present the Debiaser for Multiple Variables (DEMV), an approach able to mitigate unbalanced groups bias (i.e., bias caused by an unequal distribution of instances in the population) in both binary and multi-class classification problems with multiple sensitive variables. The proposed method is compared, under several conditions, with a set of well-established baselines using different categories of classifiers. At first we conduct a specific study to understand which is the best generation strategies and their impact on DEMV’s ability to improve fairness. Then, we evaluate our method on a heterogeneous set of datasets and we show how it overcomes the established algorithms of the literature in the multi-class classification setting and in the binary classification setting when more than two sensitive variables are involved. Finally, based on the conducted experiments, we discuss strengths and weaknesses of our method and of the other baselines.     

Writer identification approach based on bag of words with OBI features

Handwriter identification aims to simplify the task of forensic experts by providing them with semi-automated tools in order to enable them to narrow down the search to determine the final identification of an unknown handwritten sample. An identification algorithm aims to produce a list of predicted writers of the unknown handwritten sample ranked in terms of confidence measure metrics for use by the forensic expert will make the final decision.Most existing handwriter identification systems use either statistical or model-based approaches. To further improve the performances this paper proposes to deploy a combination of both approaches using Oriented Basic Image features and the concept of graphemes codebook. To reduce the resulting high dimensionality of the feature vector a Kernel Principal Component Analysis has been used. To gauge the effectiveness of the proposed method a performance analysis, using IAM dataset for English handwriting and ICFHR 2012 dataset for Arabic handwriting, has been carried out. The results obtained achieved an accuracy of 96% thus demonstrating its superiority when compared against similar techniques.     

An end-to-end deep generative approach with meta-learning optimization for zero-shot object classification

Zero-shot object classification aims to recognize the object of unseen classes whose supervised data are unavailable in the training stage. Recent zero-shot learning (ZSL) methods usually propose to generate new supervised data for unseen classes by designing various deep generative networks. In this paper, we propose an end-to-end deep generative ZSL approach that trains the data generation module and object classification module jointly, rather than separately as in the majority of existing generation-based ZSL methods. Due to the ZSL assumption that unseen data are unavailable in the training stage, the distribution of generated unseen data will shift to the distribution of seen data, and subsequently causes the projection domain shift problem. Therefore, we further design a novel meta-learning optimization model to improve the proposed generation-based ZSL approach, where the parameters initialization and the parameters update algorithm are meta-learned to assist model convergence. We evaluate the proposed approach on five standard ZSL datasets. The average accuracy increased by the proposed jointly training strategy is 2.7% and 23.0% for the standard ZSL task and generalized ZSL task respectively, and the meta-learning optimization further improves the accuracy by 5.0% and 2.1% on two ZSL tasks respectively. Experimental results demonstrate that the proposed approach has significant superiority in various ZSL tasks.     

Reliably Assessing Growth with Longitudinal Diagnostic Classification Models

Recent advances have enabled diagnostic classification models (DCMs) to accommodate longitudinal data. These longitudinal DCMs were developed to study how examinees change, or transition, between different attribute mastery statuses over time. This study examines using longitudinal DCMs as an approach to assessing growth and serves three purposes: (1) to define and evaluate two reliability measures to be used in the application of longitudinal DCMs; (2) through simulation, demonstrate that longitudinal DCM growth estimates have increased reliability compared to longitudinal item response theory models; and (3) through an empirical analysis, illustrate the practical and interpretive benefits of longitudinal DCMs. A discussion describes how longitudinal DCMs can be used as practical and reliable psychometric models when categorical and criterion‐referenced interpretations of growth are desired.     

Digital Module 05: Diagnostic Measurement—The G‐DINA Framework https://ncme.elevate.commpartners.com

In this ITEMS module, we introduce the generalized deterministic inputs, noisy “and” gate (G‐DINA) model, which is a general framework for specifying, estimating, and evaluating a wide variety of cognitive diagnosis models. The module contains a nontechnical introduction to diagnostic measurement, an introductory overview of the G‐DINA model, as well as common special cases, and a review of model‐data fit evaluation practices within this framework. We use the flexible GDINA R package, which is available for free within the R environment and provides a user‐friendly graphical interface in addition to the code‐driven layer. The digital module also contains videos of worked examples, solutions to data activity questions, curated resources, a glossary, and quizzes with diagnostic feedback.     

Ancient Mural Classification Method Based on Improved AlexNet Network

ABSTRACT

As an important part of art and culture, ancient murals depict a variety of different artistic images, and these individual images have important research value. For research purposes, it is often important to first determine the type of objects represented in a painting. However, the mural painting environment makes datasets difficult to collect, and long-term exposure leads to underlying features that are not distinct, which makes this task challenging. This study proposes a convolutional neural network model based on the classic AlexNet network model and combines it with feature fusion to automatically classify ancient mural images. Due to the lack of large-scale mural datasets, the model first expands the dataset by applying image enhancement algorithms such as scaling, brightness conversion, noise addition, and flipping; then, it extracts the underlying features (such as fresco edges) shared by the first stage of a dual channel structure. Subsequently, a second-stage deep abstraction is conducted on the features extracted by the first stage using a two-channel network, each of which has a different structure. The obtained characteristics from both channels are merged, and a loss function is constructed to obtain the classification result. This approach improves the model's robustness and feature expression ability. The model achieves an accuracy of 84.24%, a recall rate of 84.15%, and an F1-measure of 84.13% when applied to a constructed mural image dataset. Compared with the AlexNet model and other improved convolutional neural network models, the proposed model improves each evaluation index by approximately 5%, verifying the rationality and effectiveness of the model for automatic mural image classification. The mural classification model proposed in this paper comprehensively considers the influences of network width and depth and can extract rich details from mural images from multiple local channels. An effective classification method could help researchers manage and protect mural images in an orderly fashion and quickly and effectively search for target images in a digital mural library based on a specified image category, aiding mural condition monitoring and restoration efforts as well as archaeological and art historical research.     

中外情报学论文创新性特征研究

[目的/意义] 综合运用定性与定量相结合的方法对近年中外情报学论文的创新性进行分析和对比，揭示情报学领域研究的创新性特征，发现领域学术论文中创新句内部的知识关系，进行更细粒度的论文创新性分析，为研究领域创新点深层次利用提供条件，同时丰富科技论文创新性监测的途径，促进科学研究创新。[方法/过程] 从句子级创新性识别出发，选取中英文各两种情报学期刊作为样本，采用信息抽取和机器学习的方法，将创新句的抽取从现有的摘要扩展到全文，充分利用句子结构和句法特征识别领域创新内容，探讨近年中外情报学论文在创新对象、主题、类别等方面的特征，并做对比分析，最后通过对自动分类的论文集合进行定性的内容分析，总结归纳出中外情报学论文创新的表达范式。[结果/结论] 从创新的表达来看，中外情报学论文创新句的分布情况基本一致，英文期刊论文创新的表达更丰富。从创新性特征来看，英文情报学期刊论文创新主题较集中，而中文主题多样和分散；具体方法的创新是近年情报学领域的创新热点，而在研究方法上创新不足；中英文情报学期刊论文的创新性特点都反映了应用研究、实证研究的成果较多，而理论创新推动缓慢的趋势。     

基于深度学习的中文专利自动分类方法研究

[目的/意义] 面向当前国内专利审查和专利情报分析工作中对于海量专利分类的客观需求，设计了7种基于深度学习的专利自动分类方法，对比各种方法的分类效果，从而助力专利分类效率和效果的提升。[方法/过程] 针对传统机器学习方法存在的缺陷，基于Word2Vec、CNN、RNN、Attention机制等深度学习技术，考虑专利文本语序特征、上下文特征以及分类关键特征，设计Word2Vec+TextCNN、Word2Vec+GRU、Word2Vec+BiGRU、Word2Vec+BiGRU+TextCNN等7种深度学习模型，以中国专利为例，选取IPC主分类号的"部"作为分类依据，对比这7种模型与3种传统分类模型在中文专利分类任务中的效果。[结果/结论] 实证研究效果显示，采用考虑语序特征、上下文特征及强化关键特征的深度学习方法进行中文专利分类具有更优的分类效果。     

分级视角下残疾人冬季两项运动的关键竞技特征分析

残疾人冬季两项项目竞技特征的理论研究薄弱和对残疾分级体系的认知模糊是影响我国残疾人冬季两项项目发展的关键问题。基于残疾分级视角,对有关残疾人冬季两项项目竞技特征的文献进行梳理与分析,认为:1)可通过加强我国冬季两项残疾人运动员运动技术的时效性和经济性练习,提升冬季两项残疾人运动员滑雪技术的动态平衡能力和滑行过程中的有氧代谢供能水平。2)在进行射击项目训练时,通过增加脱敏训练和电子激光靶系统训练等专项训练加强我国冬季两项残疾人运动员的射击技术稳定性和动作控制准确性。3)残疾人冬季两项竞技运动要求运动员具备良好的中枢神经系统调节能力,以最短时间激活副交感神经降低心率来提高动静转换效率,通过体能分配与技术运用等专项训练来全面提高冬季两项残疾人运动员的快速应变能力。4)冬季两项专项训练还需通过兼顾与整合2个小项相结合的训练方法来提升训练的整体效果。5)同一残疾分级的坐式运动员的不同坐姿对身体关键发力部位的肌肉募集有一定影响,在规则允许范围内,通过外部辅助和内部增强技术优化运动员竞技能力结构是冬季两项残疾人运动员达到国际水平的关键。     

“流空间”视角下粤港澳大湾区空间布局分级特征研究

为加快粤港澳大湾区区域一体化进程,基于“流空间”视角,以粤港澳大湾区11个核心城市为研究对象,从客运流、经济流、信息流三个维度建立数理统计模型,分析粤港澳大湾区的交通、物流和经济信息化水平,并综合运用Jenks自然断裂点法和Moran散点图对粤港澳大湾区进行空间布局分级特征研究。研究结果表明:粤港澳大湾区有两个增长极——广州和深圳;粤港澳大湾区以珠江为界,珠江东西两岸分级明显;港珠澳大桥的连通有助于加强流要素空间互动,发挥更强的区域辐射和带动作用,扩大H-H连绵区的分布范围。