Elegant Unsupervised Cross-Modal Hashing

<div><p>Unsupervised cross-modal retrieval has received increasing attention recently, because of the extreme difficulty of labeling the explosive multimedia data. The core challenge of it is how to measure the similarities between multi-modal data without label information. In previous works, various distance metrics are selected for measuring the similarities and predicting whether samples belong to the same class. However, these predictions are not always right. Unfortunately, even a few wrong predictions can undermine the final retrieval performance. To address this problem, in this paper, we categorize predictions as solid and soft ones based on their confidence. We further categorize samples as solid and soft ones based on the predictions. We propose that these two kinds of predictions and samples should be treated differently. Besides, we find that the absolute values of similarities can represent not only the similarity but also the confidence of the predictions. Thus, we first design an elegant dot product fusion strategy to obtain effective inter-modal similarities. Subsequently, utilizing these similarities, we propose a generalized and flexible weighted loss function where larger weights are assigned to solid samples to increase the retrieval performance, and smaller weights are assigned to soft samples to decrease the disturbance of wrong predictions. Despite less information is used, empirical studies show that the proposed approach achieves the state-of-the-art retrieval performance.</p><br></div>

Elegant Unsupervised Cross-Modal Hashing

<div><p>Unsupervised cross-modal retrieval has received increasing attention recently, because of the extreme difficulty of labeling the explosive multimedia data. The core challenge of it is how to measure the similarities between multi-modal data without label information. In previous works, various distance metrics are selected for measuring the similarities and predicting whether samples belong to the same class. However, these predictions are not always right. Unfortunately, even a few wrong predictions can undermine the final retrieval performance. To address this problem, in this paper, we categorize predictions as solid and soft ones based on their confidence. We further categorize samples as solid and soft ones based on the predictions. We propose that these two kinds of predictions and samples should be treated differently. Besides, we find that the absolute values of similarities can represent not only the similarity but also the confidence of the predictions. Thus, we first design an elegant dot product fusion strategy to obtain effective inter-modal similarities. Subsequently, utilizing these similarities, we propose a generalized and flexible weighted loss function where larger weights are assigned to solid samples to increase the retrieval performance, and smaller weights are assigned to soft samples to decrease the disturbance of wrong predictions. Despite less information is used, empirical studies show that the proposed approach achieves the state-of-the-art retrieval performance.</p><br></div>

Recent development of PeakForce Tapping mode atomic force microscopy and its applications on nanoscience

Nanoscience is a booming field incorporating some of the most fundamental questions concerning structure, function, and applications. The cutting-edge research in nanoscience requires access to advanced techniques and instrumentation capable of approaching these unanswered questions. Over the past few decades, atomic force microscopy (AFM) has been developed as a powerful platform, which enables in situ characterization of topological structures, local physical properties, and even manipulating samples at nanometer scale. Currently, an imaging mode called PeakForce Tapping (PFT) has attracted more and more attention due to its advantages of nondestructive characterization, high-resolution imaging, and concurrent quantitative property mapping. In this review, the origin, principle, and advantages of PFT on nanoscience are introduced in detail. Three typical applications of this technique, including high-resolution imaging of soft samples in liquid environment, quantitative nanomechanical property mapping, and electrical/electrochemical property measurement will be reviewed comprehensively. The future trends of PFT technique development will be discussed as well.

Исследование структуры объемного гетероперехода в полимерных солнечных элементах с помощью комбинации ультрамикротомирования и атомно-силовой микроскопии

A method for visualization via atomic-force microscopy of the internal structure of photoactive layers of polymer solar cells using an ultramicrotome for photoactive layer cutting is proposed and applied. The method creates an opportunity to take advantage of atomic-force microscopy in structural investigations of the bulk of soft samples. Such advantages of atomic-force microscopy include a high contrast and the ability to measure various surface properties at nanometer resolution. Using the proposed method, samples of the photoactive layer of polymer solar cells based on a mixture of PTB7 polythiophene and PC_71BM fullerene derivatives are studied. The disclosed details of the bulk structure of this mixture allow us to draw additional conclusions about the effect of morphology on the efficiency of organic solar cells.