Embedding Mechanism Research Articles

Systemic inflammatory regulators and proliferative diabetic retinopathy: A bidirectional Mendelian randomization study.

Increasing evidence shows that systemic inflammation is an embedded mechanism of proliferative diabetic retinopathy (PDR). However, the specific systemic inflammatory factors involved in this process remained obscure. The study aimed to identify the upstream and downstream systemic regulators of PDR by using Mendelian randomization (MR) analyses. We performed a bidirectional two-sample MR analysis implementing the results from genome-wide association studies for 41 serum cytokines from 8,293 Finnish individuals, and PDR from FinnGen consortium (2,025 cases vs. 284,826 controls) and eight cohorts of European ancestry (398 cases vs. 2,848 controls), respectively. The inverse-variance-weighted method was adopted as the main MR method, and four additional MR methods (MR-Egger, weighted-median, MR-pleiotropy residual sum and outlier (MR-PRESSO), and MR-Steiger filtering methods) were used for the sensitivity analyses. Results from FinnGen and eight cohorts were pooled into a meta-analysis. Our results showed that genetically predicted higher stem cell growth factor-β (SCGFb) and interleukin-8 were positively associated with an elevated risk of PDR, with a combined effect of one standard deviation (SD) increase in SCGFb and interleukin-8 causing 11.8% [95% confidence interval (CI): 0.6%, 24.2%]) and 21.4% [95% CI: 3.8%, 41.9%]) higher risk of PDR, respectively. In contrast, genetically predisposition to PDR showed a positive association with the increased levels of growth-regulated oncogene-α (GROa), stromal cell-derived factor-1 alpha (SDF1a), monocyte chemotactic protein-3 (MCP3), granulocyte colony-stimulating factor (GCSF), interleukin-12p70, and interleukin-2 receptor subunit alpha (IL-2ra). Our MR study identified two upstream regulators and six downstream effectors of PDR, providing opportunities for new therapeutic exploitation of PDR onset. Nonetheless, these nominal associations of systemic inflammatory regulators and PDR require validation in larger cohorts.

Kyoto Protocol and Paris Agreement: Transition from Bindings to Pledges – A Review

The present thematic review assessed 47 articles on the world’s global climate management policy. It reviews the Kyoto Protocol and Paris Agreement aiming to mitigate climate change with the mechanisms of bindings and pledges, respectively. The research on Kyoto Protocol suggests that most Annex I parties have met their targets through its embedded mechanisms. However, global emissions continued to increase even in the protocol period owing to many defects in its inherent mechanism. Further, the review of the Paris Agreement reveals that most of its signatories are striving hard to achieve the targets. Despite that, the global emissions trends have been very alarming and predicted emissions could be far above the Paris Agreement’s limit. The review reveals that the Kyoto Protocol successfully paved the way for a new climate management order. At the same time, the success of the Paris Agreement relies mainly on communication and compliance of respective Nationally Determined Contributions (NDCs) by signatories, alignments of NDCs and Intended Nationally Determined Contributions (INDCs), and transparency in their emission mitigation targets. The study opined that the developed nations should help developing nations meet their pledges, transfer first-hand technology and provide adequate innovative technological solutions with financial provisions for making a carbon-neutral sustainable planet.

A Novel Reversible Data Hiding Scheme Based on Pixel-Residual Histogram

Prediction-error expansion (PEE) is the most popular reversible data hiding (RDH) technique due to its efficient capacity-distortion tradeoff. With the generated prediction-error histogram (PEH) and adaptively selected expansion bins, the image redundancy is well exploited by PEE. However, for the most widely used rhombus predictor, the rounding operation which groups different prediction-errors into one value is completely unnecessary. The embedding can be extended to a general case by removing the rounding operation, and more histogram bins can be derived for expansion with a new mapping mechanism. Therefore, in this article, instead of pixel prediction-error, we propose to compute the pixel residuals without the rounding operation, and a new embedding mechanism based on pixel-residual histogram (PRH) modification is devised. In PRH, four bins correspond to one bin in PEH. Then, different from the one-to-one mapping between the prediction-error and pixel modification, a four-to-one mapping between the pixel-residual and pixel modification is established, and the performance is optimized by adaptively selecting four expansion bin pairs for embedding. Since more modification selections are considered, better performance can be obtained. Moreover, the proposed scheme is extended to the two-dimensional (2D) histogram and multiple histograms based embedding, and the performance is further enhanced. The superiority of the proposed method is experimentally verified by comparing it with some state-of-the-art works.

TDNet: A Novel Transductive Learning Framework With Conditional Metric Embedding for Few-Shot Remote Sensing Image Scene Classification

Few-shot learning, which aims to learn the concept of novel category from extremely limited labeled samples, has received intense interests in remote sensing image scene classification. Most of the existing methods inherit the philosophy of prototype learning and tackle classification as the prototype-based metric matching problem. Despite the achievement that has been obtained so far, the problems of interclass metric misalignment and intraclass variations have become two main challenges that obstacle the performance. In this paper, a novel transductive learning framework with conditional metric embedding is proposed to remedy these problems. First, a conditional metric embedding mechanism is introduced to perform anisotropic embedding for each pair of the support category and query instance. This design provides the model with flexible scalability to accommodate the metric biases across classes. Second, a transductive prototype learning strategy is presented to enhance the robustness of the prototype against intraclass variations. The unlabeled query instances are transformed into pseudo instances equipped with credibility coefficients and then leveraged to calibrate the prototype estimation bias in low data regimes. Third, a long-term consistency regularization is designed, which continuously memorizes the historical prototypes to generate additional supervision of interclass separation in the global label space. Benefiting from this design, the discriminability of the prototypes obtains obvious improvement. Finally, extensive experiments are conducted on three public benchmark remote sensing datasets. The experimental results demonstrate the validity and the superiority of the proposed method in low-shot conditions.

ANSD-MA-MSE: Adaptive Neural Speaker Diarization Using Memory-Aware Multi-Speaker Embedding

In this paper, we propose a neural speaker diarization (NSD) network architecture consisting of three key components. First, a memory-aware multi-speaker embedding (MA-MSE) mechanism is proposed to facilitate a dynamical refinement of speaker embedding to reduce a potential data mismatch between the speaker embedding extraction and the NSD network. Next, a speaker selection procedure is introduced to handle situations where the detected number of speakers is different from the assumed speaker size in the NSD network. Finally, an adaptive procedure is proposed to improve the required prior information for the nonoverlap speech segments in a given utterance during each iteration. We call our proposed framework adaptive neural speaker diarization with memory-aware multi-speaker embedding (ANSD-MA-MSE). Our method improves diarization performance in realistic operating scenarios, such as adverse acoustic environments, domain mismatches, and a varying, rather than fixed, number of speakers. Having been tested on both the AMI corpus and the DIHARD-III evaluation sets, our proposed approach consistently outperforms other state-of-the-art techniques in diarization error rates, including the results reported by the best single-model system in the DIHARD-III challenge. Our code is publicly available at <uri xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">https://github.com/Maokui-He/NSD-MA-MSE</uri> .

Investigating the Impact of Dual Network Embedding and Dual Entrepreneurial Bricolage on Knowledge-Creation Performance: An Empirical Study in Fujian, China

Aim/Purpose: This study investigates the relationship between dual network embedding, dual entrepreneurial bricolage, and knowledge-creation performance. Background: The importance of new ventures for innovation and economic growth has been fully endorsed. Establishing incubation organizations to help new startups overcome constraints and dilemmas has become the consensus of various countries. In particular, the number of Chinese makerspaces has rapidly increased. Startups in the makerspaces form a loosely coupled dual network to cooperate and share resources, especially knowledge. Methodology: By convenience sampling, 400 startups in the makerspaces in Fujian Province, China were selected for the questionnaire survey study. In total, 307 valid responses were collected, yielding a response rate of 76.8%. The survey data were analyzed for hypothesis testing, using the PL-SEM technique with the AMOS20.0 software. Contribution: At the theoretical level, this research supplements the exploration of the influencing factors of the entrepreneurial bricolage of startups at the network level. It deepens the research on the internal mechanism of the dual network embeddedness affecting the knowledge-creation performance. In practice, it provides a theoretical basis and management inspiration for startups in makerspaces to overcome the inherent disadvantage of being too small and weak to explore innovative paths. Findings: First, relational embedding of startups in makerspaces directly affects knowledge-creation performance. Second, dual entrepreneurial bricolage plays a mediating role in diversity. Selective entrepreneurial bricolage plays a partial mediating role between relationship embedding and knowledge-creation performance. Parallel entrepreneurial bricolage plays a complete intermediary role between structural embedding and knowledge-creation performance. Dual entrepreneurial bricolage plays a complete intermediary role between knowledge embedding and knowledge-creation performance. Recommendations for Practitioners: Enterprises in the makerspaces should make dynamic adjustments to the network embedded state and dual entrepreneurial bricolage to improve knowledge-creation performance. When startups conduct selective entrepreneurship bricolage, they should strengthen relational and knowledge embeddedness to improve their relationship strength and tacit knowledge acquisition. When startups conduct parallel entrepreneurship bricolage, structural and knowledge embedding should be strengthened to improve the position of enterprises in the network to acquire diversified knowledge to explore and discover new business opportunities and project resources. Recommendation for Researchers: The heterogeneity of industries and regions may impact the dual network embedding mechanism of startups. Researchers can choose a wider range of regions and industries for sampling. Impact on Society: This study provides a theoretical basis and management inspiration for startups to overcome the inherent disadvantage of being too small and weak to explore innovative paths. It provides a basis to support startups in unleashing innovation vitality and achieving healthy growth. Future Research: Previous studies have shown that network relationships and bricolage behavior have a certain relationship with the enterprise life cycle. Future research can adopt a longitudinal research design across time points, which will increase the explanatory power of research conclusions.

Preparation and characterization of high embedding efficiency epigallocatechin-3-gallate glycosylated nanocomposites

Glycosylated protein nano encapsulation was an efficient encapsulation technology, but its embedding rate for EGCG was not high, and the research on the embedding mechanism was relatively weak. Based on this, this study compared the embedding effect of glycosylated peanut globulin and glycosylated casein on EGCG. The embedding mechanism of EGCG with glycosylated protein was discussed by ultraviolet, fluorescence, infrared and fluorescence microscopy. Results revealed that the highest encapsulation efficiency of EGCG was 93.89 ± 1.11%. The neutral pH value and 0.3 mg/mL EGCG addition amount were suitable for EGCG glycosylated nanocomposites. The hydrogen bond between EGCG hydroxyl group and tyrosine and tryptophan of glycosylated protein is mainly non covalent. The encapsulation effect of EGCG glycosylated nanocomposites could be quenched by changing the polar environment and spatial structure of the group. The fluorescence characteristic and dispersibility of EGCG glycosylated peanut globin were higher than EGCG glycosylated casein. This study might provide a theoretical basis for EGCG microencapsulation technology and EGCG application in tea beverage and liquid tea food systems.

Graph based multi-scale neighboring topology deep learning for kidney and tumor segmentation

Objective. Effective learning and modelling of spatial and semantic relations between image regions in various ranges are critical yet challenging in image segmentation tasks. Approach. We propose a novel deep graph reasoning model to learn from multi-order neighborhood topologies for volumetric image segmentation. A graph is first constructed with nodes representing image regions and graph topology to derive spatial dependencies and semantic connections across image regions. We propose a new node attribute embedding mechanism to formulate topological attributes for each image region node by performing multi-order random walks (RW) on the graph and updating neighboring topologies at different neighborhood ranges. Afterwards, multi-scale graph convolutional autoencoders are developed to extract deep multi-scale topological representations of nodes and propagate learnt knowledge along graph edges during the convolutional and optimization process. We also propose a scale-level attention module to learn the adaptive weights of topological representations at multiple scales for enhanced fusion. Finally, the enhanced topological representation and knowledge from graph reasoning are integrated with content features before feeding into the segmentation decoder. Main results. The evaluation results over public kidney and tumor CT segmentation dataset show that our model outperforms other state-of-the-art segmentation methods. Ablation studies and experiments using different convolutional neural networks backbones show the contributions of major technical innovations and generalization ability. Significance. We propose for the first time an RW-driven MCG with scale-level attention to extract semantic connections and spatial dependencies between a diverse range of regions for accurate kidney and tumor segmentation in CT volumes.

Unsupervised Deep Event Stereo for Depth Estimation

Bio-inspired event cameras have been considered effective alternatives to traditional frame-based cameras for stereo depth estimation, especially in challenging conditions such as low-light or high-speed environments. Recently, deep learning-based supervised event stereo matching methods have achieved significant performance improvements over the traditional event stereo methods. However, the supervised methods depend on ground-truth disparity maps for training, and it is difficult to secure a large amount of ground-truth disparity maps. A feasible alternative is to devise an unsupervised event stereo method that can be trained without ground-truth disparity maps. To this end, we propose the first unsupervised event stereo matching method that can predict dense disparity maps, and is trained by transforming the depth estimation problem into a warping-based reconstruction problem. We propose a novel unsupervised loss function that enforces the network to minimize the feature-level epipolar correlation difference between the ground-truth intensity images and warped images. Moreover, we propose a novel event embedding mechanism that utilizes both temporal and spatial neighboring events to capture spatio-temporal relationships among the events for stereo matching. Experimental results reveal that the proposed method outperforms the baseline unsupervised methods by significant margins (e.g., up to 16.88% improvement) and achieves comparable results with the existing supervised methods. Extensive ablation studies validate the efficacy of the proposed modules and architectural choices.

Extensive Dynamics of Global Collaboration Network for Scientific Breakthroughs

Studies on knowledge networks have paid less attention to peripheral countries, and emerging scientific linkages are particularly important for peripheral countries to obtain global knowledge. For this reason, we focus on extensive dynamics of peripheral countries in the scientific collaboration network; that is, the establishment and end of scientific links. We build a global collaboration network for scientific breakthroughs based on coauthored publications in Nature, Science, and Proceedings of the National Academy of Sciences of the United States of America (PNAS) from the Incites database, and identify the role of collaborators based on authors’ sequence. The results show that (1) extensive dynamics are dominated by peripheral countries and unevenly distributed between hierarchies; (2) both network structure and proximity factors (geographical, cognitive, institutional, and cultural proximity) take significantly positive effects on the establishment of new scientific links, with differences between incumbents and new entrants; and (3) the embedding mechanism of peripheral countries in the network can be summarized as preferential attachment rule, “two-step” strategy and “springboard” effect, with some core countries serving as a springboard.

A MULTI-PLATFORM APPROACH USING HYBRID DEEP LEARNING MODELS FOR AUTOMATIC DETECTION OF HATE SPEECH ON SOCIAL MEDIA

Hate speech on online social networks is a general problem across social media platforms that has the potential of causing physical harm to the society. The growing number of hateful comments on the Internet and the rate at which tweets and posts are published per second on social media make it a challenging task to manually identify and remove the hateful commentsfrom such posts. Although numerous publications have proposed machine learning approaches to detect hate speech and other antisocial online behaviours without concentrating on blocking the hate speech from being published on social media. Similarly, prior publications on deep learning and multi-platform approaches did not work on the topic of detecting hate speech in Englishlanguage comments on Twitter and Facebook. This paper proposed a deep learning approach based on a hybrid of convolutional neural network (CNN) and long short-term memory (LSTM) with pre-trained GloVe words embedding to automatically detect and block hate speech on multiple social media platforms including Twitter and Facebook. Thus, datasets were collected from Twitter and Facebook which were annotated as hateful and non-hateful. A set of features were extracted from the datasets based on word embedding mechanism, and the word embeddings were fed into our deep learning framework. The experiment was carried out as a three independent tasks approach. The results show that our hybrid CNN-LSTM approach in Task 1 achieved an f1-score of 0.91, Task 2 obtained an f1-score of 0.92, and Task 3 achieved an f1-score of 0.87. Thus, there is outstanding performance in classifying text as Hate speech or non-hate speech in all the considered metrics. Based on the findings, we conclude that hatespeech can be detected and blocked on social media before it can reach the public.

RMTF-Net: Residual Mix Transformer Fusion Net for 2D Brain Tumor Segmentation.

Due to the complexity of medical imaging techniques and the high heterogeneity of glioma surfaces, image segmentation of human gliomas is one of the most challenging tasks in medical image analysis. Current methods based on convolutional neural networks concentrate on feature extraction while ignoring the correlation between local and global. In this paper, we propose a residual mix transformer fusion net, namely RMTF-Net, for brain tumor segmentation. In the feature encoder, a residual mix transformer encoder including a mix transformer and a residual convolutional neural network (RCNN) is proposed. The mix transformer gives an overlapping patch embedding mechanism to cope with the loss of patch boundary information. Moreover, a parallel fusion strategy based on RCNN is utilized to obtain local–global balanced information. In the feature decoder, a global feature integration (GFI) module is applied, which can enrich the context with the global attention feature. Extensive experiments on brain tumor segmentation from LGG, BraTS2019 and BraTS2020 demonstrated that our proposed RMTF-Net is superior to existing state-of-art methods in subjective visual performance and objective evaluation.

Promoting the Pipeline of Diverse Individuals Within Gastroenterology: Reflections From AGA FORWARD Program Scholars

Promoting the Pipeline of Diverse Individuals Within Gastroenterology: Reflections From AGA FORWARD Program Scholars

Multi-Modal Alignment of Visual Question Answering Based on Multi-Hop Attention Mechanism

The alignment of information between the image and the question is of great significance in the visual question answering (VQA) task. Self-attention is commonly used to generate attention weights between image and question. These attention weights can align two modalities. Through the attention weight, the model can select the relevant area of the image to align with the question. However, when using the self-attention mechanism, the attention weight between two objects is only determined by the representation of these two objects. It ignores the influence of other objects around these two objects. This contribution proposes a novel multi-hop attention alignment method that enriches surrounding information when using self-attention to align two modalities. Simultaneously, in order to utilize position information in alignment, we also propose a position embedding mechanism. The position embedding mechanism extracts the position information of each object and implements the position embedding mechanism to align the question word with the correct position in the image. According to the experiment on the VQA2.0 dataset, our model achieves validation accuracy of 65.77%, outperforming several state-of-the-art methods. The experimental result shows that our proposed methods have better performance and effectiveness.

Improving health workforce governance: the role of multi-stakeholder coordination mechanisms and human resources for health units in ministries of health

BackgroundA cohesive and strategic governance approach is needed to improve the health workforce (HW). To achieve this, the WHO Global Strategy on Human Resources for Health (HRH) promotes mechanisms to coordinate HRH stakeholders, HRH structures and capacity within the health sector to support the development and implementation of a comprehensive HW agenda and regular reporting through WHO’s National Health Workforce Accounts (NHWA).MethodsUsing an adapted HRH governance framework for guidance and analysis, we explored the existence and operation of HRH coordination mechanisms and HRH structures in Malawi, Nepal, Sudan and additionally from a global perspective through 28 key informant interviews and a review of 165 documents.ResultsA unified approach is needed for the coordination of stakeholders who support the timely development and oversight of an appropriate costed HRH strategy subsequently implemented and monitored by an HRH unit. Multiple HRH stakeholder coordination mechanisms co-exist, but the broader, embedded mechanisms seemed more likely to support and sustain a comprehensive intersectoral HW agenda. Including all stakeholders is challenging and the private sector and civil society were noted for their absence. The credibility of coordination mechanisms increases participation. Factors contributing to credibility included: high-level leadership, organisational support and the generation and availability of timely HRH data and clear ownership by the ministry of health.HRH units were identified in two study countries and were reported to exist in many countries, but were not necessarily functional. There is a lack of specialist knowledge needed for the planning and management of the HW amongst staff in HRH units or equivalent structures, coupled with high turnover in many countries. Donor support has helped with provision of technical expertise and HRH data systems, though the benefits may not be sustained.ConclusionWhile is it important to monitor the existence of HRH coordination mechanisms and HRH structure through the NHWA, improved ‘health workforce literacy’ for both stakeholders and operational HRH staff and a deeper understanding of the operation of these functions is needed to strengthen their contribution to HW governance and ultimately, wider health goals.

Constructive Fuzzy Cognitive Map for Depression Severity Estimation.

Depression is a common and serious medical disorder that negatively affects the mood and the emotions of people, especially adolescents. In this paper, a novel framework for automatically creating Fuzzy Cognitive Maps (FCMs) is proposed. It is applied for the estimation of the severity of depression among adolescents, based on their electroencephalogram (EEG). The introduced Constructive FCM (CFCM) utilizes features extracted by a Constructive Fuzzy Representation Model (CFRM), which conduces to detect in a more intuitive way the cause-and-effect relationships between the brain activity and depression. CFCM contributes to limiting the participation of experts, and the manual interventions in the traditional construction of FCMs, it provides an embedded mechanism for dimensionality reduction, and it constitutes an inherently interpretable approach to decision making, while being uncertainty-aware and simple to implement. The results of the experiments, using a recent publicly available dataset, demonstrate the effectiveness of the proposed framework and highlight its advantages.

GAN‐based deep neural networks for graph representation learning

AbstractGraph representation learning has attracted increasing attention in a variety of applications that involve learning on non‐Euclidean data. Recently, generative adversarial networks(GAN) have been increasingly applied to the field of graph representation learning, and large progress has been made. However, most GAN‐based graph representation learning methods use adversarial learning strategies directly on the update of the vector representation instead of the deep embedding mechanism. Compared with deep models, these methods are less capable of learning nonlinear features. To address this problem and take full advantage of the essential advantages of GAN, we propose to use adversarial idea on the reconstruction mechanism of deep autoencoders. Specifically, the generator and the discriminator are the two basic components of the GAN structure. We use the deep autoencoder as the discriminator, which can capture the highly nonlinear structure of the graph. In addition, the generator another generative model is introduced into the adversarial learning system as a competitor. A series of empirical results proved the effectiveness of the new approach.

Design of a Planar Cable-Driven Parallel Crane Without Parasitic Tilt

Abstract Cable-driven parallel robots (CDPRs) offer high payload capacities, large translational workspace, and high dynamics performances. Their rotational workspace is generally far more limited, however, which can be resolved using cable loops, as was shown in previous research. In the case of fully constrained CDPRs, cable loops can induce unwanted torques on the moving-platform (MP), causing it to tilt and move away from its intended position, which we call parasitic tilt. Hence, the orientation accuracy of such robots is usually limited. This paper deals with the design, modeling, and prototyping of a planar CDPR with infinite rotations, without parasitic tilt and without an additional motor. This robot, which we call a cable-driven parallel crane (CDPC), is composed of a mobile platform (MP) with an embedded mechanism and a transmission module. The MP is linked to the frame by four cables, three of them acting in parallel, forming in effect a double parallelogram. Among these three parallel cables, two form a cable loop, i.e., they are two strands of the same cable redirected to and from the MP through an embedded pulley. The two-degree-of-freedom (dof) motions of the moving-platform of the CDPC and the internal dof of its embedded mechanism are driven by a total of three actuators, which are fixed to the frame. As a consequence, the overall system is fully actuated, its total mass and inertia in motion is reduced, and it is free of parasitic tilts.

Heterogeneous multi-scale neighbor topologies enhanced drug-disease association prediction.

Identifying new uses of approved drugs is an effective way to reduce the time and cost of drug development. Recent computational approaches for predicting drug-disease associations have integrated multi-sourced data on drugs and diseases. However, neighboring topologies of various scales in multiple heterogeneous drug-disease networks have yet to be exploited and fully integrated. We propose a novel method for drug-disease association prediction, called MGPred, used to encode and learn multi-scale neighboring topologies of drug and disease nodes and pairwise attributes from heterogeneous networks. First, we constructed three heterogeneous networks based on multiple kinds of drug similarities. Each network comprises drug and disease nodes and edges created based on node-wise similarities and associations that reflect specific topological structures. We also propose an embedding mechanism to formulate topologies that cover different ranges of neighbors. To encode the embeddings and derive multi-scale neighboring topology representations of drug and disease nodes, we propose a module based on graph convolutional autoencoders with shared parameters for each heterogeneous network. We also propose scale-level attention to obtain an adaptive fusion of informative topological representations at different scales. Finally, a learning module based on a convolutional neural network with various receptive fields is proposed to learn multi-view attribute representations of a pair of drug and disease nodes. Comprehensive experiment results demonstrate that MGPred outperforms other state-of-the-art methods in comparison to drug-related disease prediction, and the recall rates for the top-ranked candidates and case studies on five drugs further demonstrate the ability of MGPred to retrieve potential drug-disease associations.

Research on Hybrid Real-Time Picking Routing Optimization Based on Multiple Picking Stations

In the order picking process of the warehouse center, considering the rapid increase in the volume of orders arriving at the picking center at the time of the promotional festival, a hybrid operation mode with multiple picking tables is used to meet the picking requirements of the huge number of real-time orders. Therefore, in this paper, a hybrid picking mode is proposed, taking into account both the idle degree of picking stations and their order item centers of gravity, and a new reinforcement learning algorithm embedding mechanism (PRL) with placeholder control is designed to solve the problem of a huge number of real-time item orders arriving at the picking center system on promotional holidays and in inconsistent quantities, and numerical simulations are performed for this algorithm. The experimental results show that the PRL algorithm in hybrid picking mode can handle a huge number of orders simultaneously and improve picking efficiency effectively.