Dual-View Curricular Optimal Transport for Cross-Lingual Cross-Modal Retrieval.

Current research on cross-modal retrieval is mostly English-oriented, as the availability of a large number of English-oriented human-labeled vision-language corpora. In order to break the limit of non-English labeled data, cross-lingual cross-modal retrieval (CCR) has attracted increasing attention. Most CCR methods construct pseudo-parallel vision-language corpora via Machine Translation (MT) to achieve cross-lingual transfer. However, the translated sentences from MT are generally imperfect in describing the corresponding visual contents. Improperly assuming the pseudo-parallel data are correctly correlated will make the networks overfit to the noisy correspondence. Therefore, we propose Dual-view Curricular Optimal Transport (DCOT) to learn with noisy correspondence in CCR. In particular, we quantify the confidence of the sample pair correlation with optimal transport theory from both the cross-lingual and cross-modal views, and design dual-view curriculum learning to dynamically model the transportation costs according to the learning stage of the two views. Extensive experiments are conducted on two multilingual image-text datasets and one video-text dataset, and the results demonstrate the effectiveness and robustness of the proposed method. Besides, our proposed method also shows a good expansibility to cross-lingual image-text baselines and a decent generalization on out-of-domain data.

Reduction of zinc and iron in dust removal ash by H2 generated from lignite pyrolysis.

The article describes a process based on the hydrogen recovery of dust removal ash from lignite pyrolysis. The high volatile components in lignite will produce a large amount of H2 during the pyrolysis process, which can replace part of C to reduce zinc and iron oxides in dust removal ash, thereby achieving the goal of reducing carbon consumption. Two methods were used to study the different pyrolysis behaviors of lignite, and it was found that the peak of H2 release was reached at 700 °C. After 1000 °C, the pyrolysis of lignite basically ended, and the final pyrolysis gases were mainly H2, CO, CO2, CH4, and a small amount of CxHy. In addition, the proportion of H2 generated by the full pyrolysis of lignite at 1000 °C in H2-H2O is 84.4 %, and the proportion of CO in the CO-CO2 system is 74 %, which can meet the reduction of zinc and iron oxides in dust removal ash. Based on the principle of minimum free energy and laboratory experiments, zinc oxide is completely reduced to zinc vapor above 906 °C, while the Fe obtained from iron oxide reduction increases with the increase of H2/CO in the system above 810 °C, further confirming the feasibility of this process. The effects of temperature, lignite added amount and holding time on iron reduction and zinc volatilization during the reduction process of dust removal ash was investigated. The results indicate that due to the fast diffusion rate of H2, the reaction rate is greatly increased, which can quickly increase the metallization rate of dust removal ash to over 95 % and the dezincification rate also increases from 19.60 % to about 98.76 % with the increase of carbon content under low temperature (1000 °C) and low carbon content (10 wt %) conditions.

Numerical Simulation of Slag Entrainment by Vortex Flux during Tapping at Converter.

In order to improve the yield of steel produced in the converter and the quality of the molten steel, and to understand the distribution of the flow field in the converter and ladle during the steelmaking process, the CFD fluid simulation software Fluent 2020 R2 was used to analyze the flow field of the converter static steelmaking process. The aperture of the steel outlet and the timing of the vortex formation under different angles were studied, as well as the disturbance level of the injection flow in the ladle molten pool. The study revealed that in the steelmaking process, the emergence of tangential vectors caused the entrainment of slag by the vortex, whereas in the later stages of steelmaking, the turbulent flow of slag disrupted the vortex, resulting in its dissipation. When the converter angle increases to 90°, 95°, 100°, and 105°, the eddy current occurrence time is 43.55 s, 66.44 s, 68.80 s, and 72.30 s, and the eddy current stabilization time is 54.10 s, 70.36 s, 70.95 s, and 74.26 s, respectively. When the converter angle is 100-105°, it is suitable to add alloy particles into the ladle molten pool. When the tapping port diameter is 220 mm, the eddy current inside the converter changes and the mass flow rate of the tapping port is "oscillating". When the aperture of the steel outlet was 210 mm, the steelmaking time could be shortened by about 6 s without affecting the internal flow field structure of the converter.

NRG1 knockdown rescues PV interneuron GABAergic maturation deficits and schizophrenia behaviors in fetal growth restriction mice.

Schizophrenia is a highly debilitating mental disorder, those who experienced fetal growth restriction (FGR) in the early stage of life have a greater probability of schizophrenia. In this study, FGR mice showed hyperactivity in locomotor activity test, sociability dysfunction in three chamber test and nesting social behavior tests, cognition decline in Morris water maze and impaired sensory motor gating function in prepulse inhibition test. Mechanistic studies indicated that the number of parvalbumin (PV) interneuron was significantly reduced in FGR mouse media prefrontal cortex (mPFC). And the mRNA and protein level of neuregulin 1(NRG1), which is a critical schizophrenia gene, increased significantly in FGR mouse mPFC. Furthermore, NRG1 knockdown in FGR mouse mPFC improved PV interneuron GABAergic maturation and rescued schizophrenia behaviors including hyperactivity, social novelty defects, cognition decline, and sensorimotor gating deficits in FGR mice. This study indicates that mPFC NRG1 upregulation is one of the main causes of FGR-induced schizophrenia, which leads to significant reduction of PV interneuron number in mPFC. NRG1 knockdown in mPFC significantly rescues schizophrenia behaviors in FGR mouse. This study thus provides a potential effective therapy target or strategy for schizophrenia patients induced by FGR.

An Amplitude-Modulated Metadevice with Switchable Reflection, Transmission, and Absorption.

In this paper, we propose a reconfigurable metadevice with independent polarization control based on a 90° rotationally symmetric microstructure. Three functionalities of broadband high-efficiency transmission, broadband high-efficiency reflection, and perfect absorption are switched by the on-state and off-state PIN diodes. Coding metadevices designed with diversified lumped element combinations are further studied in detail. By controlling the two diodes on the top layer in opposite states, absorption bandwidth is significantly improved. Reasonable arrangements of coding sequences allow for reflected dual/multi-beam modulation. Electric field distribution, power loss, complex impedance functions, and equivalent circuit models are used to better analyze the physical mechanism of the design. A prototype of the microstructure has been fabricated, and the experimental results agree well with the simulation. Electronic components integrated microstructures with high degrees of freedom have potential applications in intelligent wireless communication, electronic detection, advanced sensors, and smart stealth radomes.

Mid-infrared reconfigurable all-dielectric metasurface based on Ge2Sb2Se4Te1 phase-change material.

In this paper, a reconfigurable all-dielectric metasurface based on Ge2Sb2Se4Te1 (GSST) phase-change material is proposed. By changing GSST from amorphous state to crystalline state, the metasurface can achieve high circular dichroism (CD) and wideband polarization conversion for circularly polarized waves in the mid-infrared (MIR) band. The maximum CD value reaches 0.95 at 74 THz, and circular polarization conversion efficiency is more than 90% in the wideband range of 41 THz - 48 THz. In addition, based on Pancharatnam-Berry phase, three kinds of wavefront manipulation of light have been realized: abnormal refraction, orbital angular momentum vortex beam and orbital angular momentum vortex beam splitting. This work has potential applications in the future MIR optical integrated system.

Video Moment Retrieval With Cross-Modal Neural Architecture Search.

The task of video moment retrieval (VMR) is to retrieve the specific video moment from an untrimmed video, according to a textual query. It is a challenging task that requires effective modeling of complex cross-modal matching relationship. Recent efforts primarily model the cross-modal interactions by hand-crafted network architectures. Despite their effectiveness, they rely heavily on expert experience to select architectures and have numerous hyperparameters that need to be carefully tuned, which significantly limit their applications in real-world scenarios. How to design flexible architectures for modeling cross-modal interactions with less manual effort is crucial for the task of VMR but has received limited attention so far. To address this issue, we present a novel VMR approach that automatically searches for an optimal architecture to learn cross-modal matching relationship. Specifically, we develop a cross-modal architecture searching method. It first searches for repeatable cell network architectures based on a directed acyclic graph, which performs operation sampling over a customized task-specific operation set. Then, we adaptively modulate the edge importance in the graph by a query-aware attention network, which performs edge sampling softly in the searched cell. Different from existing neural architecture search methods, our approach can effectively exploit the query information to reach query-conditioned architectures for modeling cross modal matching. Extensive experiments on three benchmark datasets show that our approach can not only significantly outperform the state-of-the-art methods but also run more efficiently and robustly than manually crafted network architectures.

The B7H4-PDL1 classifier stratifies immuno-phenotype in cervical cancer.

BACKGROUND: It has been revealed that B7H4 is negatively correlated with PDL1 and identifies immuno-cold tumors in glioma. However, the application of the B7H4-PDL1 classifier in cancers has not been well testified. METHODS: A pan-cancer analysis was conducted to evaluate the immunological role of B7H4 using the RNA-sequencing data downloaded from the Cancer Genome Atlas (TCGA). Immunohistochemistry (IHC) and multiplexed quantitative immunofluorescence (QIF) were performed to validate the primary results revealed by bioinformatics analysis. RESULTS: The pan-cancer analysis revealed that B7H4 was negatively correlated with PDL1 expression and immune cell infiltration in CeCa. In addition, patients with high B7H4 exhibited the shortest overall survival (OS) and relapse-free survival (RFS) while those with high PDL1 exhibited a better prognosis. Multiplexed QIF showed that B7H4 was mutually exclusive with PDL1 expression and the B7H4-high group exhibited the lowest CD8 + T cell infiltration. Besides, B7H4-high predicted highly proliferative subtypes, which expressed the highest Ki67 antigen. Moreover, B7H4-high also indicated a lower response to multiple therapies. CONCLUSIONS: Totally, the B7H4-PDL1 classifier identifies the immunogenicity and predicts proliferative subtypes and limited therapeutic options in CeCa, which may be a convenient and feasible biomarker in clinical practice.

Dual Encoding for Video Retrieval by Text.

This paper attacks the challenging problem of video retrieval by text. In such a retrieval paradigm, an end user searches for unlabeled videos by ad-hoc queries described exclusively in the form of a natural-language sentence, with no visual example provided. Given videos as sequences of frames and queries as sequences of words, an effective sequence-to-sequence cross-modal matching is crucial. To that end, the two modalities need to be first encoded into real-valued vectors and then projected into a common space. In this paper we achieve this by proposing a dual deep encoding network that encodes videos and queries into powerful dense representations of their own. Our novelty is two-fold. First, different from prior art that resorts to a specific single-level encoder, the proposed network performs multi-level encoding that represents the rich content of both modalities in a coarse-to-fine fashion. Second, different from a conventional common space learning algorithm which is either concept based or latent space based, we introduce hybrid space learning which combines the high performance of the latent space and the good interpretability of the concept space. Dual encoding is conceptually simple, practically effective and end-to-end trained with hybrid space learning. Extensive experiments on four challenging video datasets show the viability of the new method. Code and data are available at https://github.com/danieljf24/hybrid_space.

Fetal growth restriction impairs hippocampal neurogenesis and cognition via Tet1 in offspring.

Fetal growth restriction (FGR) increases the risk for impaired cognitive function later in life. However, the precise mechanisms remain elusive. Using dexamethasone-induced FGR and protein restriction-influenced FGR mouse models, we observe learning and memory deficits in adult FGR offspring. FGR induces decreased hippocampal neurogenesis from the early post-natal period to adulthood by reducing the proliferation of neural stem cells (NSCs). We further find a persistent decrease of Tet1 expression in hippocampal NSCs of FGR mice. Mechanistically, Tet1 downregulation results in hypermethylation of the Dll3 and Notch1 promoters and inhibition of Notch signaling, leading to reduced NSC proliferation. Overexpression of Tet1 activates Notch signaling, offsets the decline in neurogenesis, and enhances learning and memory abilities in FGR offspring. Our data indicate that a long-term decrease in Tet1/Notch signaling in hippocampal NSCs contributes to impaired neurogenesis following FGR and could serve as potential targets for the intervention of FGR-related cognitive disorders.

Circular dichroism metamirror with diversified chiral molecules combinations.

In this paper, we propose a giant circular dichroism (CD) chiral metamirror that differentially absorbs circularly polarized (CP) waves with dual/multi bands. The structure is composed of two variable τ resonators with counter split opening directions and different eigenfrequencies. Even more important, these two resonators are aligned vertically, and then integrated into one periodical unit cell, which results in the high-efficiency absorptive CD. The chiral metamirror has a narrow-band absorption of the left-handed circularly polarized (LCP) wave for lower resonance, and a broadband absorption of the right-handed circularly polarized (RCP) wave in the higher-frequency range. Optical resonator designs with diversified chiral molecules combinations are further studied in detail. Through the study of different periodic arrangements, parameters scanning optimization, and power loss distributions, it is proved that the overall permutation symmetry breakdown of total combined chiral molecules is the determining factor for the absorptive CD. Based on the principle in the microwave model, two mid-infrared chiral metamirrors are further realized, which can still show good spin-dependent absorption. This multi-polarization and multifunction response advance novel photonic devices for a variety of applications including CP laser, biomolecules detection, and energy harvester.

Fine-Grained Fashion Similarity Prediction by Attribute-Specific Embedding Learning.

This paper strives to predict fine-grained fashion similarity. In this similarity paradigm, one should pay more attention to the similarity in terms of a specific design/attribute between fashion items. For example, whether the collar designs of the two clothes are similar. It has potential value in many fashion related applications, such as fashion copyright protection. To this end, we propose an Attribute-Specific Embedding Network (ASEN) to jointly learn multiple attribute-specific embeddings, thus measure the fine-grained similarity in the corresponding space. The proposed ASEN is comprised of a global branch and a local branch. The global branch takes the whole image as input to extract features from a global perspective, while the local branch takes as input the zoomed-in region-of-interest (RoI) w.r.t. the specified attribute thus able to extract more fine-grained features. As the global branch and the local branch extract the features from different perspectives, they are complementary to each other. Additionally, in each branch, two attention modules, i.e., Attribute-aware Spatial Attention and Attribute-aware Channel Attention, are integrated to make ASEN be able to locate the related regions and capture the essential patterns under the guidance of the specified attribute, thus make the learned attribute-specific embeddings better reflect the fine-grained similarity. Extensive experiments on three fashion-related datasets, i.e., FashionAI, DARN, and DeepFashion, show the effectiveness of ASEN for fine-grained fashion similarity prediction and its potential for fashion reranking. Code and data are available at https://github.com/maryeon/asenpp.

Predicting scalar coupling constants by graph angle-attention neural network.

Scalar coupling constant (SCC), directly measured by nuclear magnetic resonance (NMR) spectroscopy, is a key parameter for molecular structure analysis, and widely used to predict unknown molecular structure. Restricted by the high cost of NMR experiments, it is impossible to measure the SCC of unknown molecules on a large scale. Using density functional theory (DFT) to theoretically calculate the SCC of molecules is incredibly challenging, due to the cost of substantial computational time and space. Graph neural networks (GNN) of artificial intelligence (AI) have great potential in constructing molecul ar-like topology models, which endows them the ability to rapidly predict SCC through data-driven machine learning methods, and avoiding time-consuming quantum chemical calculations. With a priori knowledge of angles, we propose a graph angle-attention neural network (GAANN) model to predict SCC by means of some easily accessible related information. GAANN, with a multilayer message-passing network and a self-attention mechanism, can accurately simulate the molecular-like topological structure and predict molecular properties. Our simulations show that the prediction accuracy by GAANN, with the log(MAE) = -2.52, is close to that by DFT calculations. Different from conventional AI methods, GAANN combining the AI method with quantum chemistry theory (Karplus equation) has a strong physicochemical interpretability about angles. From an AI perspective, we find that bond angle has the highest correlation with the SCC among all angle features (dihedral angle, bond angle, geometric angles) about multiple coupling types in the small molecule datasets.

Fetal growth restriction mice are more likely to exhibit depression-like behaviors due to stress-induced loss of dopaminergic neurons in the VTA.

Fetal growth restriction (FGR) is a severe perinatal complication that can increase risk for mental illness. To investigate the mechanism by which FGR mice develop mental illness in adulthood, we established the FGR mouse model and the FGR mice did not display obvious depression-like behaviors, but after environmental stress exposure, FGR mice were more likely to exhibit depression-like behaviors than control mice. Moreover, FGR mice had significantly fewer dopaminergic neurons in the ventral tegmental area but no difference in serotoninergic neurons in the dorsal raphe. RNA-seq analysis showed that the downregulated genes in the midbrain of FGR mice were associated with many mental diseases and were especially involved in the regulation of NMDA-selective glutamate receptor (NMDAR) activity. Furthermore, the NMDAR antagonist memantine can relieve the stress-induced depression-like behaviors of FGR mice. In summary, our findings provide a theoretical basis for future research and treatment of FGR-related depression.

Can docetaxel combined prednisone effectively treat hormone refractory prostate cancer?: A protocol of systematic review and meta-analysis.

BACKGROUND: Previous studies have reported that docetaxel combined prednisone (DP) has been used for the treatment of patients with hormone refractory prostate cancer (HRPC). However, its results are still inconsistent. Therefore, this study will synthesize the latest evidence of the efficacy and safety of DP for the treatment of patients with HRPC. METHODS: Cochrane Library, PUBMED, EMBASE, Web of Science, CINAHL, CBM, and CNKI will be searched to identify randomized controlled trials published from their inception to the March 1, 2020, irrespective language and publication time restrictions. We will calculate the pooled effects of dichotomous outcomes as risk ratio and 95% confidence intervals, and that of continuous outcomes as standardized mean difference or mean difference and 95% confidence intervals. Study quality will be assessed using Cochrane risk of bias, and quality of evidence for main outcome will be evaluated using Grading of Recommendations Assessment Development and Evaluation. Statistical analysis will be performed using RevMan 5.3 software. RESULTS: This study will appraise the efficacy and safety of DP for the treatment of patients with HRPC. The primary outcome includes overall survival, and the secondary outcomes comprise of progression-free survival, prostate-specific antigen response rate, duration of prostate-specific antigen response, objective tumor response rate, disease-free survival, quality of life, and adverse events. CONCLUSION: The results of this study may provide helpful evidence of DP for the treatment of patients with HRPC.Systematic review registration: INPLASY202040112.

MicroRNA-153 improves the neurogenesis of neural stem cells and enhances the cognitive ability of aged mice through the notch signaling pathway.

Aging-related cognitive ability impairments are one of the main threats to public health, and impaired hippocampal neurogenesis is a major cause of cognitive decline during aging. However, the regulation of adult neurogenesis in the hippocampus requires further study. Here, we investigated the role of microRNA-153 (miR-153), a highly conserved microRNA in mice and humans, in adult neurogenesis. During the passaging of neural stem cells (NSCs) in vitro, endogenous miR-153 expression was downregulated, with a decrease in neuronal differentiation ability. In addition, miR-153 overexpression increased the neurogenesis of NSCs. Further studies showed that miR-153 regulated neurogenesis by precisely targeting the Notch signaling pathway through inhibition of Jagged1 and Hey2 translation. In vivo analysis demonstrated that miR-153 expression was decreased in the hippocampi of aged mice with impaired cognitive ability, and that miR-153 overexpression in the hippocampus promoted neurogenesis and markedly increased the cognitive abilities of the aged mice. Overall, our findings revealed that miR-153 affected neurogenesis by regulating the Notch signaling pathway and elucidated the function of miR-153 in aging-related, hippocampus-dependent cognitive ability impairments, and neurodegenerative diseases.

Chirality selective metamaterial absorber with dual bands.

In this paper, a sensitive chirality selective metamaterial absorber (CSMA) is constructed by using 'I-shaped' resonator with asymmetric twisted metallic wires. Absorption of 95.18% and 91.77% at two resonant frequencies can be achieved for left-handed circularly polarized (LCP) incident wave, with little loss of right-handed circularly polarized (RCP) incident wave, which results in significant absorptive circular dichroism. Not only can the CSMA intensely absorb LCP illumination with dual bands, but also circularly polarized (CP) conversion for RCP wave is achieved over a broad bandwidth. The spin-dependent absorption, closely linked to chiral symmetry breaking, is investigated through oblique incidence, power loss distribution and scanning parameters optimization. The proposed strategy is further demonstrated in mid-infrared band which could advance the applications in polarization manipulation to circularly polarized detectors/lasers, chiral sensing/bolometers, and molecular spectroscopy.

Trifunctional metasurfaces: concept and characterizations.

Achieving multiple diversified functionalities in a single flat device is crucial for electromagnetic (EM) integration. While many recent efforts were devoted to designing multifunctional metasurfaces, most meta-devices realized so far typically exhibit only two functionalities. In this paper, we propose a generic strategy to design trifunctional metasurfaces, based on carefully designed single structure meta-atoms possessing polarization-controlled transmission/reflection properties. As a proof of our concept, we design and fabricate a trifunctional metasurface possessing simultaneously three distinct functionalities including beam splitting, deflecting, and focusing, and perform both far-field and near-field microwave experiments to demonstrate the predicted functionalities of the fabricated device. Experimental results are in good agreement with numerical simulations. These findings can motivate the realizations of high-performance multifunctional meta-devices in different frequency domains and with diversified functionalities.

Tim-3-expressing macrophages are functionally suppressed and expanded in oral squamous cell carcinoma due to virus-induced Gal-9 expression.

Oropharyngeal head and neck squamous cell carcinoma is a common malignant tumor in the oral cavity. High-risk human papillomavirus 16 infection is a major cause of oropharyngeal head and neck squamous cell carcinoma development. Strong antitumor immune responses, especially CD8+ T cell responses, are thought to be essential to effective cancer treatment and are associated with better prognosis in oropharyngeal head and neck squamous cell carcinoma. In this study, we examined the role of the Tim-3/Gal-9 pathway in oropharyngeal head and neck squamous cell carcinoma patients. We found that Gal-9 expression by CD4+ T cells was increased in human papillomavirus-positive oropharyngeal head and neck squamous cell carcinoma patients, but not in human papillomavirus-negative oropharyngeal head and neck squamous cell carcinoma patients. Increased Gal-9 secretion by CD4+ T cells presented multiple immunosuppressive effects. Coculturing monocytes with high Gal-9-expressing CD4+ T cells resulted in the expansion of Tim-3+ monocytes, which suppressed interferon gamma production by activated CD8+ T cells. Subsequently, total monocytes incubated with exogenous Gal-9, or high Gal-9-expressing CD4+ T cells, suppressed the expression of interferon gamma by CD8+ T cells. Exogenous Gal-9 and high Gal-9-expressing CD4+ T cells also suppressed the secretion of both interleukin 10 and interleukin 12 by monocytes. These effects are Tim-3/Gal-9-dependent because blocking Tim-3 and/or Gal-9 could enhance the support of CD8+ T cell interferon gamma production and the interleukin 10 and interleukin 12 secretion by monocytes. Together, these data suggest that the high Tim-3 expression in monocytes could be utilized by tumor-promoting Gal-9 expression on CD4+ T cells. Immunotherapy in human papillomavirus-positive oropharyngeal head and neck squamous cell carcinoma patients therefore faces an additional challenge posed by Tim-3 and Gal-9 and likely requires the blockade of these molecules.

[Construction of three HLA-A*0201-peptide tetramers and their preliminary application in detection of vitiligo-specific cytotoxic T lymphocytes].

OBJECTIVE: To construct vitiligo-specific HLA-A*0201-peptide tetramers and to apply the constructed tetramers in detection of vitiligo-specific cytotoxic T lymphocytes (CTL). METHODS: Proteins HLA-A0201*-BSP and ß2M were obtained by effective prokaryotic expression. The purified proteins were refolded with vitiligo antigen peptides MelanA 26-35, gp100 209-217, and tyrosinase 1-9, respectively to form HLA-A*0201-peptide complex. The complex was biotinylated by BirA enzyme and purified by gel-filtration chromatography. The tetramers were generated by mixing the complex with phycoerythrin (PE)-streptavidin at a ratio of 4∶1 and identified by Dot-blot assay. The capacity of tetramer to detect vitiligo-specific CTL was analyzed by flow cytometry. RESULTS: The biotinylation of vitiligo-specific HLA-A*0201-peptide tetramers were successfully performed by Dot-blot. Flow cytometry analysis indicated that the tetramer effectively bound to specific CTL from peripheral blood of patients with vitiligo. CONCLUSION: Three kinds of biotinylated vitiligo-specific HLA-A*0201-peptide tetramers have been constructed successfully. The tetramer can detect antigen specific CTL from patients with vitiligo.