Learning discriminative and generalizable features with multi-branch for person re-identification

Finer-grained local features play a supplementary role in the description of pedestrian global features, and the combination of them has been an essential solution to improve discriminative performances in person re-identification (PReID) tasks. The existing part-based methods mostly extract representational semantic parts according to human visual habits or some prior knowledge and focus on spatial partition strategies but ignore the significant influence of channel information on PReID task. So, we proposed an end-to-end multi-branch network architecture (MCSN) jointing multi-level global fusion features, channel features and spatial features in this paper to better learn more diverse and discriminative pedestrian features. It is worth noting that the effect of multi-level fusion features on the performance of the model is taken into account when extracting global features. In addition, to enhance the stability of model training and the generalization ability of the model, the BNNeck and the joint loss function strategy are applied to all vector representation branches. Extensive comparative evaluations are conducted on three mainstream image-based evaluation protocols, including Market-1501, DukeMTMC-ReID and MSMT17, to validate the advantages of our proposed model, which outperforms previous state-of-the-art in ReID tasks.

Discovery of genes that modulate flavivirus replication in an interferon-dependent manner

Establishment of the interferon (IFN)-mediated antiviral state provides a crucial initial line of defense against viral infection. Numerous genes that contribute to this antiviral state remain to be identified. Using a loss-of-function strategy, we screened an original library of 1156 siRNAs targeting 386 individual curated human genes in stimulated microglial cells infected with Zika virus (ZIKV), an emerging RNA virus that belongs to the flavivirus genus. The screen recovered twenty-one potential host proteins that modulate ZIKV replication in an IFN-dependent manner, including the previously known IFITM3 and LY6E. Further characterization contributed to delineate the spectrum of action of these genes towards other pathogenic RNA viruses, including Hepatitis C virus and SARS-CoV-2. Our data revealed that APOL3 acts as a proviral factor for ZIKV and several other related and unrelated RNA viruses. In addition, we showed that MTA2, a chromatin remodeling factor, possesses potent flavivirus-specific antiviral functions. Our work identified previously unrecognized genes that modulate the replication of RNA viruses in an IFN-dependent way, opening new perspectives to target weakness points in the life cycle of these viruses.

LncRNATUG1 Facilitates Th2 Cell Differentiation by Targeting the miR-29c/B7-H3 Axis on Macrophages

The role of long non-coding RNAs (lncRNA) in asthma remains unclear. In this study, we examined the role of long non-coding RNA taurine upregulated 1 (lncRNA TUG1) in asthma. We found that lncRNA TUG1 is one of the differentially expressed lncRNAs in the monocytes of asthmatic children and is associated with Th cell differentiation. LncRNA TUG1 and miR-29c are mainly distributed in the cytoplasm of macrophages. Our data suggested that lncRNA TUG1 increased in macrophages stimulated by House Dust Mite in a dose-dependent manner. Using loss- and gain of function strategy, we found that miR-29c might regulate Th2 cell differentiation by directly targeting co-stimulatory molecule B7-H3. Furthermore, down-regulation of lncRNA TUG1 decreased the level of GATA3 in CD4+T cells and was associated with miR-29c/B7-H3 axis. Moreover, the dual-luciferase reporter assay confirmed that lncRNA TUG1 serves as a competing endogenous RNA to sponge miR-29c. According to the rescue experiment, lncRNA TUG1 regulated Th2 cell differentiation via miR-29c. These data suggest that lncRNA TUG1 in macrophages regulates Th2 cell differentiation via miR-29c/B7-H3 axis.

Goodrich et al. Embedded Executive Function Strategy Instruction

Prior research indicates that training aspects of executive function (EF; e.g., working memory) does not generalize beyond tasks that are closely aligned with the training (e.g., Melby-Lervåg & Hulme, 2013). However, such evaluations of training EF to improve performance in related areas of cognitive development (e.g., academic skills) have not utilized EF training paradigms in the context of evidence-based academic skills instruction. Therefore, the purpose of this pilot study was to evaluate the potential for embedding EF training within evidence-based early literacy instruction for young, at-risk dual language learners (DLLs). Sixty-nine preschool DLLs were randomly assigned to one of three treatment groups: a business-as-usual control group (BAU), a group that received early literacy instruction only (EL group), and a group that received early literacy instruction with embedded EF training (EL+EF group). The EL+EF group significantly outperformed the BAU control group for two outcomes (vocabulary and syntax skills), and significantly outperformed the EL group on one outcome (syntax). Interaction effects suggested that early literacy instruction, regardless of embedded EF supports, was most effective for children with low pretest EF. In contrast, the added benefits of embedded EF training beyond the effects of early literacy instruction alone were largest for children with high pretest EF. Implications for academic-cognitive hybrid interventions are discussed.

The Lysine Specific Demethylase-1 Negatively Regulates the COL9A1 Gene in Human Articular Chondrocytes

Osteoarthritis (OA) is a degenerative disease of the joints which is associated with an impaired production of the cartilage matrix by the chondrocytes. Here, we investigated the role of Lysine-Specific Demethylase-1 (LSD1), a chromatin remodeling enzyme whose role in articular chondrocytes was previously associated with a catabolic activity and which is potentially involved during OA. Following a loss of function strategy and RNA sequencing analysis, we detail the genes which are targeted by LSD1 in human articular chondrocytes and identify COL9A1, a gene encoding the α1 chain of the cartilage-specific type IX collagen, as negatively regulated by LSD1. We show that LSD1 interacts with the transcription factor SOX9 and is recruited to the promoter of COL9A1. Interestingly, we observe that OA cartilage displays stronger LSD1 immunostaining compared with normal, and we demonstrate that the depletion of LSD1 in OA chondrocytes prevents the decrease in COL9A1 following Il-1β treatment. These results suggest LSD1 is a new regulator of the anabolic activity of articular chondrocytes potentially destabilizing the cartilage matrix, since it negatively regulates COL9A1, a gene encoding a crucial anchoring collagen molecule. This newly identified role played by LSD1 may thus participate in the alteration of the cartilage matrix during OA.

Systematic Reconstruction of the Complete Two-Component Sensorial Network in Staphylococcus aureus

ABSTRACT In bacteria, adaptation to changes in the environment is mainly controlled through two-component signal transduction systems (TCSs). Most bacteria contain dozens of TCSs, each of them responsible for sensing a different range of signals and controlling the expression of a repertoire of target genes (regulon). Over the years, identification of the regulon controlled by each individual TCS in different bacteria has been a recurrent question. However, limitations associated with the classical approaches used have left our knowledge far from complete. In this report, using a pioneering approach in which a strain devoid of the complete nonessential TCS network was systematically complemented with the constitutively active form of each response regulator, we have reconstituted the regulon of each TCS of S. aureus in the absence of interference between members of the family. Transcriptome sequencing (RNA-Seq) and proteomics allowed us to determine the size, complexity, and insulation of each regulon and to identify the genes regulated exclusively by one or many TCSs. This gain-of-function strategy provides the first description of the complete TCS regulon in a living cell, which we expect will be useful to understand the pathobiology of this important pathogen. IMPORTANCE Bacteria are able to sense environmental conditions and respond accordingly. Their sensorial system relies on pairs of sensory and regulatory proteins, known as two-component systems (TCSs). The majority of bacteria contain dozens of TCSs, each of them responsible for sensing and responding to a different range of signals. Traditionally, the function of each TCS has been determined by analyzing the changes in gene expression caused by the absence of individual TCSs. Here, we used a bacterial strain deprived of the complete TC sensorial system to introduce, one by one, the active form of every TCS. This gain-of-function strategy allowed us to identify the changes in gene expression conferred by each TCS without interference of other members of the family.

Modified King’s Family for Multiple Zeros of Scalar Nonlinear Functions

There is no doubt that there is plethora of optimal fourth-order iterative approaches available to estimate the simple zeros of nonlinear functions. We can extend these method/methods for multiple zeros but the main issue is to preserve the same convergence order. Therefore, numerous optimal and non-optimal modifications have been introduced in the literature to preserve the order of convergence. Such count of methods that can estimate the multiple zeros are limited in the scientific literature. With this point, a new optimal fourth-order scheme is presented for multiple zeros with known multiplicity. The proposed scheme is based on the weight function strategy involving functions in ratio. Moreover, the scheme is optimal as it satisfies the hypothesis of Kung–Traub conjecture. An exhaustive study of the convergence is shown to determine the fourth order of the methods under certain conditions. To demonstrate the validity and appropriateness for the proposed family, several numerical experiments have been performed. The numerical comparison highlights the effectiveness of scheme in terms of accuracy, stability, and CPU time.

Automatic clustering using quantum-based multi-objective emperor penguin optimizer and its applications to image segmentation

In this study, a novel quantum-based multi-objective is proposed using Schrödinger equations. The two new operations namely weighted cluster centroid computation and threshold setting are also introduced to refine cluster centroids. A novel fitness function strategy is also proposed for efficient searching. The proposed technique is compared with various well-known approaches. Experimental outcomes show that the proposed quantum approach outperforms other existing approaches.