Data Homogeneity Effect in Deep Learning-Based Prediction of Type 1 Diabetic Retinopathy

This study is aimed at evaluating a deep transfer learning-based model for identifying diabetic retinopathy (DR) that was trained using a dataset with high variability and predominant type 2 diabetes (T2D) and comparing model performance with that in patients with type 1 diabetes (T1D). The Kaggle dataset, which is a publicly available dataset, was divided into training and testing Kaggle datasets. In the comparison dataset, we collected retinal fundus images of T1D patients at Chang Gung Memorial Hospital in Taiwan from 2013 to 2020, and the images were divided into training and testing T1D datasets. The model was developed using 4 different convolutional neural networks (Inception-V3, DenseNet-121, VGG1, and Xception). The model performance in predicting DR was evaluated using testing images from each dataset, and area under the curve (AUC), sensitivity, and specificity were calculated. The model trained using the Kaggle dataset had an average (range) AUC of 0.74 (0.03) and 0.87 (0.01) in the testing Kaggle and T1D datasets, respectively. The model trained using the T1D dataset had an AUC of 0.88 (0.03), which decreased to 0.57 (0.02) in the testing Kaggle dataset. Heatmaps showed that the model focused on retinal hemorrhage, vessels, and exudation to predict DR. In wrong prediction images, artifacts and low-image quality affected model performance. The model developed with the high variability and T2D predominant dataset could be applied to T1D patients. Dataset homogeneity could affect the performance, trainability, and generalization of the model.

Do White People See Variation in Black Skin Tones? Reexamining a Purported Outgroup Homogeneity Effect

Previous research has reported that white survey interviewers remember black respondents’ skin tones in a much narrower range than recollections by black interviewers. This finding has been used to suggest that, in line with the one-drop rule, whites do not perceive meaningful differences between light- and dark-skinned black people. The authors reanalyze evidence thought to demonstrate relative homogeneity in white interviewers’ evaluation of black skin tones. In contrast to previous studies, this examination of several data sources reveals significant heterogeneity in the ratings assigned by white interviewers when taking into account the ordinal nature of the skin tone measures. The results are consistent with theories of social cognition that emphasize that beyond formal racial classification schemes, skin tone is used to implicitly categorize others along a continuum of “blackness.” The findings also align with research suggesting that rather than nullifying within-race skin tone, increases in white racism intensify white colorism.

Human face-selective cortex does not distinguish between members of a racial outgroup

People often fail to individuate members of social outgroups, a phenomenon known as the outgroup homogeneity effect. Here, we used fMRI repetition suppression to investigate the neural representation underlying this effect. In a pre-registered study, White human perceivers (N = 29) responded to pairs of faces depicting White or Black targets. In each pair, the second face depicted either the same target as the first face, a different target from the same race, or a scrambled face outline. We localized face-selective neural regions via an independent task, and demonstrated that neural activity in the fusiform face area distinguished different faces only when targets belonged to the perceivers’ racial ingroup (White). By contrast, face-selective cortex did not discriminate between other-race individuals. Moreover, across two studies (total N = 67) perceivers were slower to discriminate between different outgroup members and remembered them to a lesser extent. Together, these results suggest that the outgroup homogeneity effect arises when early-to-mid-level visual processing results in an erroneous overlap of representations of outgroup members.

Research of grain size homogeneity effect on sheet stamping ability characteristics of Al2Mg and Al6Mg alloys

The quality of products made of sheet aluminum alloys strongly depends on the technological features of the sheet stamping process, as well as on the structure of sheet semi-finished products. The grain size and grain structure uniformity are among the key structural features that influence stampability. A method is proposed and the homogeneity of the grain structure is evaluated. Stampability of Al2Mg and Al6Mg aluminium alloys was evaluated based on measurements of the spring back index, minimum bending radius, stamping ratio, and Martens strain index. Cold work (with a strain degree of 20 %) and subsequent recrystallization annealing at temperatures of 250, 350 and 450 °C for 1 h were used to obtain a grain structure of (26,8 Ѓ} 7,4)÷(126 Ѓ} 43) μm (Al6Mg alloy) and (120 Ѓ} 11)÷(264 Ѓ} 130) μm (Al2Mg alloy) in size. As a result of processing, the effect of the initial grain size was revealed: the coarser structure of the Al2Mg alloy led to a larger grain size after strain and annealing. It was found that an increase in the grain size in both alloys leads to an increase in the Martens index and a decrease in the stamping ratio, which indicates higher stampability of the alloys in the drawing operations of sheet stamping. In the Al2Mg alloy, an increase in the grain size leads to a decrease in the spring back index by 1,5–1,7 times, and an increase in the minimum bending radius. In the Al6Mg alloy, an increase in the grain size leads to an increase in the spring back index by 1,1–1,2 times, and a decrease in the minimum bending radius. The Al6Mg minimum bending radius remains higher compared to Al2Mg regardless of the grain size. Grain size inhomogeneity in the Al6Mg alloy causes an increase in the Martens index and minimum bending radius, and a decrease in the stamping ratio. In the Al2Mg alloy, grain size inhomogeneity causes an increase in the Martens index and minimum bending radius, and a decrease in the stamping ratio. For the spring back index, the increase in grain size inhomogeneity causes a high scatter of data. In the Al6Mg alloy, the low annealing temperature led to the preservation of the non-recrystallized structure, which influenced the decrease in stampability.

Topology: A Theory of a Pseudometric-Based Clustering Model and Its Application in Content-Based Image Retrieval

The clustering problem has been extensively studied over the last 50 years; however, it still has the attention of researchers. This paper presents a topological basis of a pseudometric-based clustering model which takes into account the local and global topological properties of the data to be clustered, as per the definition of homogeneity measurement. The proposed approach takes into account the homogeneity effect produced when a new particle is added to a group. The additional element can be accumulated in the group if its local homogeneity is not altered and, therefore, it is not necessary to carry out tests in another group. A new group needs to be generated if the threshold of the local homogeneity of the group exceeds. Theoretical results, their implementation, and their application to the problem of Content Based Image Retrieval (CBIR) are presented. The tests were performed using three image databases widely used in the literature, which are “Vogel and Shiele,” “Oliva and Torralba,” and “L. Fei- Fei, R. Fergus and P. Perona.” The results are presented and compared with the most competitive methods available in the literature.

Institutional Determinants of Bilateral Trade Flows: A Panel Data Analysis

This study analyzes the institutional determinants of bilateral trade flows and homogeneity effect for Pakistan with ECO countries by using panel data for years 2003-2014. Gravity trade model is estimated through panel least squares technique. Impact of institutions is very important for international trade as international businesses involve many governance systems. The results show that average impact of institutional quality and bilateral trade flows is positive. Moreover, institutional homogeneity effect exhibits that bilateral trade flows are positively related with the governance similarity. Thus, institutional quality and institutional homogeneity has dominant impact on the bilateral trade flows.

In-Group Versus Out-Group Source Memory

A common finding in the source monitoring literature is that greater similarity impairs source discriminability. Experiments traditionally manipulate similarity overtly by describing or showing sources with explicitly differentiable features. However, people may also infer source characteristics themselves, which should also affect discriminability. Two studies examined inferred source characteristics by capitalizing on the out-group homogeneity effect, whereby in-group members are conceptualized as more diverse than out-group members. Participants learned about two sources who were described only as members of an in-group or an out-group and whose actions did not have higher a priori association with either group. Source memory was superior when participants believed the sources to be in-group members. This demonstrates that people spontaneously include inferred features with source representations and can capitalize on these features during source monitoring. Interestingly, information suggesting membership in one’s in-group improved performance even for sources who had previously been considered out-group members (Experiment 2).