Search

Soil is the largest carbon (C) pool in terrestrial ecosystems. A small change of soil organic carbon (SOC) storage may have a substantial effect on the CO2 concentration in the atmosphere, potentially leading to global climate change. Forest stand density has been reported to influence SOC storage, yet the effects are often inconsistent. In order to reveal the mechanisms of effect of stand density on SOC storage, larch plantations with three different stand densities (which were 2000, 3300 and 4400 trees per hectare) were chosen. Soil properties were measured in three soil layers which are: 0–20 cm, 20–40 cm and 40–60 cm. An incubation experiment with 14C-labeled cellulose addition was subsequently conducted to study the decomposition of SOC and cellulose, as well as the enzymes activity involved in C and nutrients cycle. The results showed that SOC storage increased with increasing stand density in larch plantations, which was due to the higher C stored in heavy fraction instead of light fraction in higher density. The decomposition of added cellulose decreased with increasing stand density in each soil layer, as well as the cumulative soil derived CO2 emission rate. The activity of enzymes involved in C-cycle and C- and nitrogen (N)-cycle remained unaffected by stand density in the 0–20 cm and 20–40 cm layers. The enzyme activity involved in the phosphorus (P)-cycle did not change corresponding to the stand density in each soil layer. Enzymes involved in the N-cycle showed the highest activity in the middle stand density in 0–20 cm, but no difference was observed among different densities in the subsurface layer except for tyr in the 40–60 cm layer, which showed the lowest activity in high stand density. Cellulose addition stimulated the extracellular enzymes activity involved in the C-cycle and P-cycle in the 0–20 cm layer, and the stimulation declined with increasing stand density. However, significant stimulation of cellulose addition to C-cycle involved enzymes activity was not found in the subsurface layer. We aim to reveal the mechanism of effects of stand density of larch plantations on SOC storage by focusing on the cellulose and SOC decomposition and the corresponding extracellular enzymes activity. In the plots of higher stand density, larch plantations may lead to a weaker C output and stronger C input, which leads to the higher SOC storage.

Multi-agent reinforcement learning (MARL) has achieved great progress in cooperative tasks in recent years. However, in the local reward scheme, where only local rewards for each agent are given without global rewards shared by all the agents, traditional MARL algorithms lack sufficient consideration of agents' mutual influence. In cooperative tasks, agents' mutual influence is especially important since agents are supposed to coordinate to achieve better performance. In this paper, we propose a novel algorithm Mutual-Help-based MARL (MH-MARL) to instruct agents to help each other in order to promote cooperation. MH-MARL utilizes an expected action module to generate expected other agents' actions for each particular agent. Then, the expected actions are delivered to other agents for selective imitation during training. Experimental results show that MH-MARL improves the performance of MARL both in success rate and cumulative reward., Comment: Accepted by 2023 IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP 2023)

Hybrid beamforming is widely recognized as an important technique for millimeter wave (mmWave) multiple input multiple output (MIMO) systems. Generalized spatial modulation (GSM) is further introduced to improve the spectrum efficiency. However, most of the existing works on beamforming assume the perfect channel state information (CSI), which is unrealistic in practical systems. In this paper, joint optimization of downlink pilot training, channel estimation, CSI feedback, and hybrid beamforming is considered in GSM aided frequency division duplexing (FDD) mmWave MIMO systems. With the help of deep learning, the GSM hybrid beamformers are designed via unsupervised learning in an end-to-end way. Experiments show that the proposed multi-resolution network named GsmEFBNet can reach a better achievable rate with fewer feedback bits compared with the conventional algorithm., 4 pages, 4 figures. This work has been submitted to the IEEE for possible publication. Copyright may be transferred without notice

Communication in multi-agent reinforcement learning has been drawing attention recently for its significant role in cooperation. However, multi-agent systems may suffer from limitations on communication resources and thus need efficient communication techniques in real-world scenarios. According to the Shannon-Hartley theorem, messages to be transmitted reliably in worse channels require lower entropy. Therefore, we aim to reduce message entropy in multi-agent communication. A fundamental challenge is that the gradients of entropy are either 0 or infinity, disabling gradient-based methods. To handle it, we propose a pseudo gradient descent scheme, which reduces entropy by adjusting the distributions of messages wisely. We conduct experiments on two base communication frameworks with six environment settings and find that our scheme can reduce message entropy by up to 90% with nearly no loss of cooperation performance., Accpeted by ICC 2023

Session-based recommendation (SBR) problem, which focuses on next-item prediction for anonymous users, has received increasingly more attention from researchers. Existing graph-based SBR methods all lack the ability to differentiate between sessions with the same last item, and suffer from severe popularity bias. Inspired by nowadays emerging contrastive learning methods, this paper presents a Simple Contrastive Graph Neural Network for Session-based Recommendation (SimCGNN). In SimCGNN, we first obtain normalized session embeddings on constructed session graphs. We next construct positive and negative samples of the sessions by two forward propagation and a novel negative sample selection strategy, and then calculate the constructive loss. Finally, session embeddings are used to give prediction. Extensive experiments conducted on two real-word datasets show our SimCGNN achieves a significant improvement over state-of-the-art methods.

Over these years, multi-agent reinforcement learning has achieved remarkable performance in multi-agent planning and scheduling tasks. It typically follows the self-play setting, where agents are trained by playing with a fixed group of agents. However, in the face of zero-shot coordination, where an agent must coordinate with unseen partners, self-play agents may fail. Several methods have been proposed to handle this problem, but they either take a lot of time or lack generalizability. In this paper, we firstly reveal an important phenomenon: the zero-shot coordination performance is strongly linearly correlated with the similarity between an agent's training partner and testing partner. Inspired by it, we put forward a Similarity-Based Robust Training (SBRT) scheme that improves agents' zero-shot coordination performance by disturbing their partners' actions during training according to a pre-defined policy similarity value. To validate its effectiveness, we apply our scheme to three multi-agent reinforcement learning frameworks and achieve better performance compared with previous methods., Comment: Accepted by ICAPS 2023

Camera localization is a classical computer vision task that serves various Artificial Intelligence and Robotics applications. With the rapid developments of Deep Neural Networks (DNNs), end-to-end visual localization methods are prosperous in recent years. In this work, we focus on the scene coordinate prediction ones and propose a network architecture named as Structure Guidance Learning (SGL) which utilizes the receptive branch and the structure branch to extract both high-level and low-level features to estimate the 3D coordinates. We design a confidence strategy to refine and filter the predicted 3D observations, which enables us to estimate the camera poses by employing the Perspective-n-Point (PnP) with RANSAC. In the training part, we design the Bundle Adjustment trainer to help the network fit the scenes better. Comparisons with some state-of-the-art (SOTA) methods and sufficient ablation experiments confirm the validity of our proposed architecture.

Channel state information (CSI) feedback is necessary for the frequency division duplexing (FDD) multiple input multiple output (MIMO) systems due to the channel non-reciprocity. With the help of deep learning, many works have succeeded in rebuilding the compressed ideal CSI for massive MIMO. However, simple CSI reconstruction is of limited practicality since the channel estimation and the targeted beamforming design are not considered. In this paper, a jointly optimized network is introduced for channel estimation and feedback so that a spectral-efficient beamformer can be learned. Moreover, the deployment-friendly subarray hybrid beamforming architecture is applied and a practical lightweight end-to-end network is specially designed. Experiments show that the proposed network is over 10 times lighter at the resource-sensitive user equipment compared with the previous state-of-the-art method with only a minor performance loss., Comment: 4 pages, 5 figures, 1 table. This work has been submitted to the IEEE for possible publication. Copyright may be transferred without notice

The area of transfer learning comprises supervised machine learning methods that cope with the issue when the training and testing data have different input feature spaces or distributions. In this work, we propose a novel transfer learning algorithm called Renewing Iterative Self-labeling Domain Adaptation (Re-ISDA). In this work, we propose a novel transfer learning algorithm called Renewing Iterative Self-labeling Domain Adaptation (Re-ISDA).

Visual localization is a fundamental task that regresses the 6 Degree Of Freedom (6DoF) poses with image features in order to serve the high precision localization requests in many robotics applications. Degenerate conditions like motion blur, illumination changes and environment variations place great challenges in this task. Fusion with additional information, such as sequential information and Inertial Measurement Unit (IMU) inputs, would greatly assist such problems. In this paper, we present an efficient client-server visual localization architecture that fuses global and local pose estimations to realize promising precision and efficiency. We include additional geometry hints in mapping and global pose regressing modules to improve the measurement quality. A loosely coupled fusion policy is adopted to leverage the computation complexity and accuracy. We conduct the evaluations on two typical open-source benchmarks, 4Seasons and OpenLORIS. Quantitative results prove that our framework has competitive performance with respect to other state-of-the-art visual localization solutions., Submitted to ICRA 2023

Emerging evidence has suggested the detrimental role of oxidative stress in aggravating ischemia and reperfusion (IR) injury in diabetic livers. Interplay between oxidative stress and mitophagy has been shown. However, the role and mechanism of mitophagy in regulating oxidative stress and IR injury in diabetic livers remain unclear.Wild-type and db/db (DB) mice were subjected to a partial warm liver IR model. Liver injury, oxidative stress, mitophagy and related molecular pathways were analyzed.Here, we found that increased liver IR injury was observed in DB mice, as evidenced by higher levels of serum alanine aminotransferase and serum aspartate, worsened liver architecture damage and more hepatocellular death. DB mice also showed increased mitochondrial oxidative stress. Mitochondrial reactive oxygen species scavenge alleviated liver IR injury in DB mice. Mechanistic analysis showed that 5' adenosine monophosphate-activated protein kinase-mediated mitophagy was suppressed in DB mice post-IR. Pharmacological activation of 5' adenosine monophosphate-activated protein kinase by its agonist effectively restored mitophagy activation, leading to decreased mitochondrial oxidative stress and attenuated liver IR injury in DB mice.Our findings showed that diabetes increased oxidative stress to exacerbate liver IR injury by impairing 5' adenosine monophosphate-activated protein kinase-mediated mitophagy. Strategies targeting oxidative stress and mitophagy might provide a promising approach to ameliorate liver IR injury in diabetes patients.

Based on analysis of the formation mechanism and characteristics of the negative friction in collapsible loess areas, this study investigates the load transfer law of a wall-soil system under a vertical load, establishes the vertical bearing model of a lattice diaphragm wall, and analyzes the vertical bearing capacity of an assembled latticed diaphragm wall (ALDW) in a loess area. The factors influencing the vertical bearing characteristics of the ALDW in a loess area are analyzed. The vertical bearing mechanism of the lattice diaphragm wall in the loess area is investigated. The failure modes of the ALDW in the loess area are mainly shear failure of the soil around the wall and failure of the wall-soil interface. In the generation and development of negative friction, there is always a point where the relative displacement of the wall-soil interface is zero at a certain depth below the ground; at this point, the wall and soil are relative to each other. The collapsibility of loess, settlement of the wall and surrounding soil, and rate and method of immersion are the factors affecting the lattice diaphragm wall. The conclusions of this study provide a reference for the design and construction of ALDWs in loess areas.

Both silicone gel and quercetin are effective in scar treatment but have different action mechanisms. Quercetin is mainly applied in the gel form and can lead to poor adhesion of silicone gel sheet; therefore, they cannot be combined in clinical use. In this study, a silicone gel sheet that releases quercetin in a sustained manner for 48 hours was successfully developed. Four round scars (Ø: 1 cm) were made in the ears of New Zealand albino rabbits (n = 10). After scar healing, the rabbits were divided into four groups: blank control group with no treatment, silicone gel sheet group with dressing change every 2 days, quercetin group with dressing change three times daily, and combination treatment group with dressing change every 2 days. Scar assessment was performed 3 months later. Transepidermal water loss showed no difference between the combination treatment group and the silicone gel sheet group, but was lower than that in the quercetin group and the blank control group. Immunohistochemistry of CD 31 and proliferating cell nuclear antigen showed the following results: combination treatment group < silicone gel sheet group = quercetin group < blank control group. Polymerase chain reaction results showed that the expression of type-I and type-III collagen in the combination treatment group and the quercetin group was significantly lower than that in the other two groups. Thus, quercetin-modified silicone gel sheet combines the advantages of the two treatments and is more effective at inhibiting cell proliferation in scar tissue than either of the two treatments alone.

Based on the UMC 40 nm CMOS process, a programmable gain amplifier(PGA) with DC offset cancellation circuit(DCOC) is designed. The PGA adopts a closed-loop resistance negative feedback structure and consists of a cascade of two gain units. DCOC circuit is based on the traditional DC negative feedback structure, a nested feedback method is proposed to reduce the power consumption and area of DCOC circuit. The simulation results show that within the gain variation control range of 0～52 dB, the high pass cutoff frequency and the relative inhibition degree of DCOC are constant at 10 kHz and 50 dB,and the maximum correctable input misalignment at 0 dB is 110 mV. Compared with the traditional design method, the area of DCOC has been reduced by almost half.

The efficient selective and directional coupling of light to waveguiding circuitry at the nanoscale is one of the key challenges innanophotonics, as it constitutes a prerequisite for many applications, including information processing, routing, and quantumtechnologies. Various exotic nanostructures and nanoparticle arrangements have been designed to achieve directional coupling withcompact on-chip integration remaining one of the foremost hurdles to realizing many real-world devices. At the same time, selective coupling to one of several neighboring waveguides is much more difficult to achieve and control. To address this challenge, we demonstrate asubwavelength selective coupler integrated in a waveguide network, with selectivity controlled by wavelength, polarization, and angle ofincidence. We utilize a Janus source, which is composed of a superposition of electric and magnetic dipoles, supported by a silicon nanocylinder. By placing the nanocylinder between identical single mode silicon waveguides, we successfully achieve selective coupling with a highcontrast ratio between the waveguides. The operating wavelength of the Janus dipolar source can be easily tailored, and the coupling efficiency is also shown to be conveniently boosted by the addition of multiple nanocylinders. Our compact approach provides a direct pathtoward on-chip highly directional nanoscale sources for a plethora of applications, including information routing, metrology, and quantumtechnologies.

Learning to coordinate is a daunting problem in multi-agent reinforcement learning (MARL). Previous works have explored it from many facets, including cognition between agents, credit assignment, communication, expert demonstration, etc. However, less attention were paid to agents' decision structure and the hierarchy of coordination. In this paper, we explore the spatiotemporal structure of agents' decisions and consider the hierarchy of coordination from the perspective of multilevel emergence dynamics, based on which a novel approach, Learning to Advise and Learning from Advice (LALA), is proposed to improve MARL. Specifically, by distinguishing the hierarchy of coordination, we propose to enhance decision coordination at meso level with an advisor and leverage a policy discriminator to advise agents' learning at micro level. The advisor learns to aggregate decision information in both spatial and temporal domains and generates coordinated decisions by employing a spatiotemporal dual graph convolutional neural network with a task-oriented objective function. Each agent learns from the advice via a policy generative adversarial learning method where a discriminator distinguishes between the policies of the agent and the advisor and boosts both of them based on its judgement. Experimental results indicate the advantage of LALA over baseline approaches in terms of both learning efficiency and coordination capability. Coordination mechanism is investigated from the perspective of multilevel emergence dynamics and mutual information point of view, which provides a novel perspective and method to analyze and improve MARL algorithms.

The ionic-complementary self-assembling peptides discovered by Zhang Shuguang have solution-to-gel (sol-gel) transition capacity and one such peptide RADA16 has been commercialized into hemostatic agents. However, their sol-gel transition ability was not obvious because the peptide aqueous solution with a concentration greater than 1% w/v appeared to be thick and viscous. The current report describes PP-type self-assembling peptides. In addition to the ionic-complementary sequence, they have prolines at both ends of the sequence. This feature has led to better solubility, lower viscosity of the peptide solution, and simplified synthesis and purification processes while maintaining the great gelling performance of the ionic-complementary peptides. The PP-type peptides self-assembled into a well-organized nanofiber scaffold as shown by TEM. Among the PP-type peptides, the PRVDP9 sequence peptide was tested as a hemostatic agent and a mucosal elevating agent. The results were comparable to the classic RADA16. The PP-type self-assembling peptides have superior sol-gel transition ability. Therefore, it is predicted that they will be more suitable to be transported through catheters or endoscopes and have higher commercialization potential as compared with the classic self-assembling peptide sequences.

Lameness has a considerable influence on the welfare and health of dairy cows. Many attempts have been made to develop automatic lameness detection systems using computer vision technology. However, these detection methods are easily affected by the characteristics of individual cows, resulting in inaccurate detection of lameness. Therefore, this study explores an individualized lameness detection method for dairy cattle based on the supporting phase using computer vision. This approach is applied to eliminate the influence of the characteristics of individual cows and to detect lame cows and lame hooves. In this paper, the correlation coefficient between lameness and the supporting phase is calculated, a lameness detection algorithm based on the supporting phase is proposed, and the accuracy of the algorithm is verified. Additionally, the reliability of this method using computer vision technology is verified based on deep learning. One hundred naturally walking cows are selected from video data for analysis. The results show that the correlation between lameness and the supporting phase was 0.864; 96% of cows were correctly classified, and 93% of lame hooves were correctly detected using the supporting phase-based lameness detection algorithm. The mean average precision is 87.0%, and the number of frames per second is 83.3 when the Receptive Field Block Net Single Shot Detector deep learning network was used to detect the locations of cow hooves in the video. The results show that the supporting phase-based lameness detection method proposed in this paper can be used for the detection and classification of cow lameness and the detection of lame hooves with high accuracy. This approach eliminates the influence of individual cow characteristics and could be integrated into an automatic detection system and widely applied for the detection of cow lameness.

Sialolithiasis is the common disease of salivary glands. The majority of sialoliths occur in the submandibular gland or the Wharton’s duct. The length of sialolith usually measures from 1 to

This study aimed to improve the conventional rat burn wound model and to validate its utility. In total, 60 Sprague-Dawley rats were divided equally into the control and experimental groups. Altogether, 60 burn wound models with zones of stasis were created in each group. Gross visual assessments of the burn wounds were performed at 0 h, 24 h, and 48 h after burn creation. The rates of necrosis in the zones of stasis were calculated, and the blood flow from the wounds was examined. Wound tissues were collected 48 h after the burn and subjected to hematoxylin and eosin staining to determine whether the models were successfully established. The model success rates were calculated. The success rate of the burn wound models was significantly different between the control group and the experimental group (93.33% [56/60] vs. 100%; P=0.042). The Cronbach's alpha values and the respective correlation coefficients indicated that the stability of the zones of stasis in the models on the 2 sides of the spine was higher in the experimental group than in the control group. The standard deviations of the rate of necrosis, blood flow, and density of necrotic cells and apoptosis cell density, and inflammatory factor content in the zones of stasis were smaller in the experimental group than in the control group at 48 h after model construction. This suggested that the stability of repeated procedures was higher in the experimental group than in the control group. The novel device for creating burns in animal models facilitated the effective creation of zones of stasis for rat burn wound models. Both the model success rate and stability were higher compared with the conventional model construction method. In addition, the use of the novel device can better align with the requirements of self-controlled studies. This article is protected by copyright. All rights reserved.

Flocking control is a significant problem in multi-agent systems such as multi-agent unmanned aerial vehicles and multi-agent autonomous underwater vehicles, which enhances the cooperativity and safety of agents. In contrast to traditional methods, multi-agent reinforcement learning (MARL) solves the problem of flocking control more flexibly. However, methods based on MARL suffer from sample inefficiency, since they require a huge number of experiences to be collected from interactions between agents and the environment. We propose a novel method Pretraining with Demonstrations for MARL (PwD-MARL), which can utilize non-expert demonstrations collected in advance with traditional methods to pretrain agents. During the process of pretraining, agents learn policies from demonstrations by MARL and behavior cloning simultaneously, and are prevented from overfitting demonstrations. By pretraining with non-expert demonstrations, PwD-MARL improves sample efficiency in the process of online MARL with a warm start. Experiments show that PwD-MARL improves sample efficiency and policy performance in the problem of flocking control, even with bad or few demonstrations., Comment: Accepted by IEEE Vehicular Technology Conference (VTC) 2022-Fall

Additional file 1: Table S1. Cut-off values used in different studies and the number of positive and negative samples in each study. Table S2. List of descriptors calculated using Mordred. Table S3. The performance of the consensus model on the training set and each fold in the fivefold cross validation when the cutoff is 50%. Table S4. The performance of the consensus model on the training set and each fold in fivefold cross validation when the cutoff is 20%. Figure S1. The importance matrix plot for the RF model 1 when the cutoff is 50%. Figure S2. The importance matrix plot for the RF model 2 when the cutoff is 50%. Figure S3. The importance matrix plot for the RF model 3 when the cutoff is 50%. Figure S4. The importance matrix plot for the RF model 4 when the cutoff is 50%. Figure S5. The importance matrix plot for the RF model 5 when the cutoff is 50%. Figure S6. SHAP dependence plot of the top 20 features of the RF model 1 when the cutoff is 50%. Figure S7. SHAP dependence plot of the top 20 features of the RF model 2 when the cutoff is 50%. Figure S8. SHAP dependence plot of the top 20 features of the RF model 3 when the cutoff is 50%. Figure S9. SHAP dependence plot of the top 20 features of the RF model 4 when the cutoff is 50%. Figure S10. SHAP dependence plot of the top 20 features of the RF model 5 when the cutoff is 50%. Figure S11. The importance matrix plot for the RF model 1 when the cutoff is 20%. Figure S12. The importance matrix plot for the RF model 2 when the cutoff is 20%. Figure S13. The importance matrix plot for the RF model 3 when the cutoff is 20%. Figure S14. The importance matrix plot for the RF model 4 when the cutoff is 20%. Figure S15. The importance matrix plot for the RF model 5 when the cutoff is 20%. Figure S16. SHAP dependence plot of the top 20 features of the RF model 1 when the cutoff is 20%. Figure S17. SHAP dependence plot of the top 20 features of the RF model 2 when the cutoff is 20%. Figure S18. SHAP dependence plot of the top 20 features of the RF model 3 when the cutoff is 20%. Figure S19. SHAP dependence plot of the top 20 features of the RF model 4 when the cutoff is 20%. Figure S20. SHAP dependence plot of the top 20 features of the RF model 5 when the cutoff is 20%. Figure S21. (A) Importance matrix plot of the consensus model when the cutoff is 20%. (B) A statistical graph of the number of occurrences of the top 20 features that affect all models when the cutoff is 20%. Figure S22. SHAP dependence plot of the top 20 features of the consensus model when the cutoff is 20%.

In massive multiple-input multiple-output (MIMO) systems, the user equipment (UE) needs to feed the channel state information (CSI) back to the base station (BS) for the following beamforming. But the large scale of antennas in massive MIMO systems causes huge feedback overhead. Deep learning (DL) based methods can compress the CSI at the UE and recover it at the BS, which reduces the feedback cost significantly. But the compressed CSI must be quantized into bit streams for transmission. In this paper, we propose an adaptor-assisted quantization strategy for bit-level DL-based CSI feedback. First, we design a network-aided adaptor and an advanced training scheme to adaptively improve the quantization and reconstruction accuracy. Moreover, for easy practical employment, we introduce the expert knowledge of data distribution and propose a pluggable and cost-free adaptor scheme. Experiments show that compared with the state-of-the-art feedback quantization method, this adaptor-aided quantization strategy can achieve better quantization accuracy and reconstruction performance with less or no additional cost. The open-source codes are available at https://github.com/zhang-xd18/QCRNet., Comment: 9 pages, 8 figures, 5 tables. This work has been submitted to the IEEE for possible publication. Copyright may be transferred without notice

Liver cancer is one of the most common malignant diseases in the world. Segmentation and labeling of liver tumors and blood vessels in CT images can provide convenience for doctors in liver tumor diagnosis and surgical intervention. In the past decades, many state-of-the-art medical image segmentation algorithms appeared during this period. With the development of embedded devices, embedded deployment for medical segmentation and automatic reconstruction brings prospects for future automated surgical tasks. Yet, most of the existing segmentation methods mostly care about the spatial feature context and have a perception defect in the semantic relevance of medical images, which significantly affects the segmentation accuracy of liver tumors and blood vessels. Deploying large and complex models into embedded devices requires a reasonable trade-off between model accuracy, reasoning speed and model capacity. Given these problems, we introduce a multi-scale feature fusion network called TransFusionNet based on Transformer. This network achieved very competitive performance for liver vessel and liver tumor segmentation tasks, meanwhile it can improve the recognition of morphologic margins of liver tumors by exploiting the global information of CT images. Experiments show that in vessel segmentation task TransFusionNet achieved mean Dice coefficients of 0.899 and in liver tumor segmentation task TransFusionNet achieved mean Dice coefficients of 0.961. Compared with the state-of-the-art framework, our model achieves the best segmentation result. In addition, we deployed the model into an embedded micro-structure and constructed an integrated model for liver tumor vascular segmentation and reconstruction. This proprietary structure will be the exclusive component of the future medical field., 11 pages, 6 Figures

Mercury from recycled oxy-fuel gas needs to be removed because it can damage the aluminum devices in the system. In this work, low-cost NH4Br-modified rice husk char (RHCBr) was prepared as the bio...

Current wound scaffold dressing constructs can facilitate wound healing but do not exhibit antibacterial activity, resulting in high infection rates. We aimed to endow wound scaffold dressing with anti-infective ability by polyhexamethylenebiguanide (PHMB). We prepared PHMB hydrogel at varying concentrations (0.25%, 0.5%, 1%, 2%) and assessed release and cytotoxicity. PHMB hydrogel was added to the wound scaffold dressing to generate a PHMB hydrogel-modified wound scaffold dressing. Wound healing and infection prevention were evaluated using a full-thickness skin defect model in rats. In vitro, the hydrogel PHMB release time positively correlated with PHMB concentration, with 1% allowing sufficiently long release time to encompass the high-incidence period (3-5 days) of infection following wound scaffold dressing implantation. Implantation of 1% PHMB hydrogel into the skin did not cause adverse responses. in vitro cytotoxicity assays showed the PHMB hydrogel-modified wound scaffold dressing did not significantly affect proliferation of fibroblasts or vascular endothelial cells, 99.90% vs 99.84% for fibroblasts and 100.21% vs 99.28% for vascular endothelial cells at 21 days. Transplantation of PHMB hydrogel-modified wound scaffold dressing/unmodified wound scaffold dressing on the non-infected wounds of rats yielded no significant difference in relative vascularization rate, 47.40 vs 50.87 per view at 21 days, whereas bacterial content of the wound tissue in the PHMB hydrogel-modified wound scaffold dressing group was significantly lower than the unmodified wound scaffold dressing group, (1.80 ± 0.35) × 103 vs (9.34 ± 0.45) × 103 at 14 days. Prevalence of persistent wound infection in the rats receiving PHMB hydrogel-modified wound scaffold dressing transplantation onto infected wounds was significantly lower than the unmodified wound scaffold dressing group, 30% vs 100%. PHMB hydrogel-modified wound scaffold dressing exhibited suitable antibacterial ability, and its biological activity did not significantly differ from that of the unmodified wound scaffold dressing, thereby allowing it to effectively prevent infection following wound scaffold dressing implantation.

The abundant geothermal energy in hot dry rock (HDR) formations is an attractive renewable energy resource with great potential. China will develop its first HDR geothermal formation in the Gonghe Basin. HDR is a hard and low-permeability granite containing very few fluids. Development requires fluids to cyclically flow between injection and production wells to extract geothermal energy in the artificial heat transfer zone. Hydraulic fracturing is the main technology for creating flow paths. But few studies have investigated fractures in HDR geothermal formations. This paper investigated fractures as flow paths in HDR geothermal formations during hydraulic fracturing. Hydraulic fractures were simulated using a custom true-triaxial hydraulic fracturing test system in a realistic formation environment, in which a scaled wellbore was used that was built in outcrop granite rock from the Gonghe Basin. Fracture creation in granite was investigated via experiments, as well as influence factors, and what experience could be achieved. This study can be used to design and evaluate hydraulic fracturing projects in potential HDR geothermal formations.

Chlordecone (CD) is one of the common persistent organic pollutants in nature and has a profound impact on the environment and on public health. Accumulating evidence has demonstrated that neonatal exposure of CD influences adult physiology and behavior due to its estrogenic properties. Using socially monogamous mandarin voles as an experimental animal model, the present study aimed to evaluate the impact of neonatal exposure to CD on female social behaviors and central estrogen receptor alpha (ERα) expression in adulthood. After receiving a single subcutaneous injection with sesame seed oil (female control group), 17 beta-estradiol (E2 group), or CD group on postnatal Day 1, the social behaviors of adult animals and ERα expression in specific brain regions were assessed. The data indicated that CD or E2-treated female animals displayed increased affiliative behaviors and decreased aggressive behaviors with regard to the unfamiliar females in the social interaction test. In addition, CD or E2-treated female voles exhibited significant preferences to females over males in the sexual preference test. Moreover, CD-treated female animals exhibited higher levels of ERα expression in the bed nucleus of the stria terminalis, the central amygdala, the medial amygdala and the medial preoptic area compared with those of the control voles. The results suggested that neonatal exposure to CD may masculinize female social behaviors, possibly via CD-induced changes in the ERα expression of relevant brain regions.

Research purpose: To understand the behavioral factors affecting the success of college students’ Entrepreneurship. Research tools: the College Students’ Daily Success Behavior Scale (CSDSB) and the College Students’ Entrepreneurship Scale (CSES). Research methods: Psychometric method and Interview method. Research objects or samples: 32 college students in Guangdong Province who are starting their own businesses. Results: (1) The total score of college students’ daily success behavior and its five dimensions are positively correlated with the total score of college students’ entrepreneurship and its four dimensions. (2) The total score of college students’ daily success behavior and the dimension of “Excelsior (Ex)” have significant positive predictive effects on their entrepreneurial ability (F = 32.375, P < 0.001). (3) Successful entrepreneurs possess behavioral quality consistent with the dimensions of the college students’ daily behavior scale. Conclusion: Daily behavioral factors have a significant positive predictive effect on college students’ entrepreneurship.

Xinran Ma,* Ling Li,* Lei Zhang, Xiaorui Fu, Xin Li, Xinhua Wang, Jingjing Wu, Zhenchang Sun, Xudong Zhang, Xiaoyan Feng, Yu Chang, Zhiyuan Zhou, Feifei Nan, Jieming Zhang, Zhaoming Li, Mingzhi Zhang Department of Oncology, The First Affiliated Hospital of Zhengzhou University, Lymphoma Diagnosis and Treatment Center of Henan Province, Zhengzhou, Henan, People’s Republic of China*These authors contributed equally to this workCorrespondence: Mingzhi ZhangDepartment of Oncology, The First Affiliated Hospital of Zhengzhou University, Lymphoma Diagnosis and Treatment Centre of Henan Province, Zhengzhou 450002, Henan, People’s Republic of ChinaTel +86-13838565629Email mingzhi_zhang1@163.comPurpose: Treatment options for relapsed or refractory diffuse large B-cell lymphoma (RR DLBCL) represent an unmet medical need. Apatinib is a new oral tyrosine kinase inhibitor mainly targeting vascular endothelial growth factor receptor-2 (VEGFR-2) to inhibit tumour angiogenesis. In the present study, we evaluated the efficacy and safety of apatinib for patients with RR DLBCL.Patients and Methods: In this phase II, open-label, single-arm, prospective study, we enrolled patients aged 14– 70 years with treatment failure of at least two chemotherapeutic regimens using Simon’s two-stage design. All patients were administered apatinib at an initial dose of 500 mg on a 4-week cycle at home and visited the outpatient clinic every two cycles to evaluate efficacy and to record adverse events. We considered objective response rate (ORR) as the primary end point, and progression-free survival (PFS), and overall survival (OS) plus duration of response (DoR) as the secondary end point. (This trial was registered at ClinicalTrials.gov, identifier: NCT03376958.).Results: From January 2017 to February 2019, we screened 35 patients and enrolled 32 eligible patients. At the cutoff point (April 2019), we noted 2 (6.3%) complete responses, 12 (37.5%) partial responses, and 9 (28.1%) stable diseases, attributing to an ORR of 43.8% and a disease control rate of 71.9%. The median PFS and OS were 6.9 (95% confidence interval [CI], 5.8– 7.9) and 7.9 months (95% CI, 7.0– 8.7), respectively. The median DoR was 5.0 months (95% CI, 3.5– 6.5) for patients who achieved PR. The most common grade 3– 4 adverse events (AE) were hypertension (12.6%), hand–foot syndrome (9.4%), and leucopenia (6.3%). No apatinib-related deaths were noted.Conclusion: Home administration of apatinib shows promising efficacy and manageable AEs in patients with RR DLBCL.Keywords: apatinib, relapsed or refractory diffuse large B-cell lymphoma, VEGFR-2, efficacy, safety

To improve the efficiency of the wind power generation, this article proposes an active disturbance rejection controller (ADRC) based MPPT strategy, and establishes a LabVIEW FPGA platform based hardware-in-the-loop (HIL) test system. Firstly, the configuration, operation principle, mathematical model and maximum power point tracking (MPPT) control strategy of the wind power generation are analyzed. According to the angular speed tracking motion equation of the wind turbine, the ADRC based MPPT strategy is designed. Secondly, the real-time simulation model of wind speed, wind turbine and PMSG, as well as MPPT rapid control prototype (RCP) are developed by using PXI-FPGA architecture on the LabVIEW RT real-time operation platform. Then, by building a power converter and connecting it to the PXI-FPGA real-time operation platform, the HIL test system is established. Finally, the HIL real-time simulation tests are conducted with various wind speed conditions. The results not only prove the correctness of ADRC based MPPT strategy, but also demonstrate that the HIL system is an efficient tool for application in wind power generation development.

The aerodynamic characteristics of Darrieus vertical axis wind turbines (VAWTs) are affected by several geometrical parameters. Airfoil shape is one of the important factors which have not been received enough attention in the past, compared to other parameters such as solidity, number of blades, chord length, rotor diameter, pitch angle and aspect ratio. In this paper, four airfoils with varying maximum thickness (12%, 15%, 18% and 21%) are investigated by calculating the aerodynamic characteristics of Darrieus VAWTs using computational fluid dynamics technology. The power and torque characteristics, as well as their flow field characteristics, are analyzed. The results indicate that the power coefficient follows a trend of CP_NACA 0018 > CP_NACA 0015 > CP_NACA 0021 > CP_NACA 0012 below an optimum TSR, while it increases with the decrease in airfoil maximum thickness beyond the optimum TSR. The optimum TSR and operational zone also increase with the decrease in airfoil maximum thickness. An optimized airfoil thickness is determined to be between 15 and 18% for a relatively high power coefficient and appropriate optimum TSR, as well as a wider operational zone and a high efficiency band. Moreover, the airfoil maximum thickness influences the strength and region of the vorticity, as well as the interactions between blades and shed vortex, being the main reason for the observed differences in instantaneous torque coefficient. The vorticity of NACA 0015 model is weaker and smaller than that of NACA 0021 model when TSR is 2.0.

Xiangli Liu, YuFeng Shang, Ling Li, Xudong Zhang, Zhaoming Li, Mingzhi Zhang Department of Oncology, The First Affiliated Hospital of Zhengzhou University, Zhengzhou, Henan 450000, People’s Republic of ChinaCorrespondence: Mingzhi ZhangDepartment of Oncology, The First Affiliated Hospital of Zhengzhou University, Zhengzhou, Henan 450000, People’s Republic of ChinaTel +86 1 383 856 5629Email mingzhi_zhang1@163.comBackground: Peripheral T-cell lymphoma (PTCL) is often prone to relapse and progression even after formal first-line treatment, and there is no standard regimen for second-line treatment. What is more, the activity of thalidomide against this type of lymphoma is unknown.Purpose: The objective of this study was to evaluate the efficacy and safety of GDPT regimen in the treatment of relapsed/refractory peripheral T-cell lymphoma.Patients and methods: In this retrospective study, gemcitabine, cisplatin, prednisone, and thalidomide (GDPT) combination regimen was used as salvage protocol for PTCL that failed in first-line treatment for 29 patients and it was scheduled to give 6 cycles of GDPT therapy in order to better evaluate the efficacy unless there was evidence of disease progression, unacceptable toxicities, or refusal by the patient.Results: After a total of 106 cycles of GDPT regimen were administered, the result showed that the disease control rate (DCR) achieved 82.8% and overall response rate (ORR) reached 69.0% with 34.5% complete remission (CR) and 34.5% partial remission (PR). The median progression-free survival (PFS) was 10.0 months (95% CI 6.6–13.4) and median OS was 28.0 months (95% CI 19.2–36.8). And the 1-year PFS rate and 1-year OS rate were 43.6% and 64.6%, respectively. Both hematologic and non-hematologic toxicities were moderate and well tolerated. There was no treatment-related death.Conclusion: Thalidomide in combination with gemcitabine, cisplatin, prednisone regimen is a new promising approach to treating patients with relapse and refractory PTCL.Keywords: peripheral T-cell lymphoma, salvage chemotherapy, thalidomide, gemcitabine

Background: There have been few studies on large-scale samples regarding natural killer/T-cell lymphoma (NKTCL) in children and adolescents for the low incidence of NKTCL in this age group. The clinical, histopathological and molecular features of 22 young people with NKTCL were evaluated. The aim of the present study was to provide a better understanding of the clinicopathological aspects of this rare entity of lymphoma.Results: There were 15 male and 7 female patients (male/female, 2.14:1) with a median age of 15 years. The results revealed that the nasal site was the most involved region while non-nasal sites were observed in 27.3% out of all cases, and exhibited histopathological and immunohistochemical features similar to the cases in adults. Seven (36.8%) out of all the 19 patients during the follow-up died of the disease, and the median follow‑up period was 44 months. Moreover, patients treated with radiotherapy/chemotherapy showed significantly inferior OS compared with the untreated patients. High mutation frequencies were detected including KMT2C (5/5), MST1 (5/5), HLA-A (3/5) and BCL11A (3/5), which involved in modifications, tumor suppression and immune surveillance. Conclusion: NKTCL in children and adolescents is a rare type of T/NK cell neoplasms. This disease is difficult to distinguish from other EBV-related disorders in young people. Furthermore, active treatment is necessary after the diagnosis of NKTCL is confirmed.

Background: Epigenetic regulation plays vital roles in the oncogenesis and treatment of diffuse large B-cell lymphoma (DLBCL). The H3K9me3-specific histone methyltransferase SUV39H1 is an epigenetic gene that promotes the progression of a variety of malignancies. However, the roles of SUV39H1 in DLBCL remain unclear.Methods: Initially, the Oncomine, Cancer Cell Line Encyclopedia (CCLE), UALCAN and Gene Expression Profiling Interactive Analysis (GEPIA) databases were searched to explore the expression of SUV39H1 in DLBCL. The clinical parameters and pathological sections of 61 successive patients, including 47 cases of DLBCL and 14 cases of reactive lymphoid hyperplasia, were collected from January 2019 to November 2020. Immunohistochemistry was conducted to verify the results of the database search. Finally, relevant parameters and pathological results were combined to analyze the expression of SUV39H1.Results: We found that the expression of SUV39H1 in DLBCL tissues was higher than that in normal and other cancer tissues (PP=0.037), suggesting that patients with high SUV39H1 expression are less likely to develop tumors over 7.5 cm in size. Regarding survival, the group with high SUV39H1 expression had a lower survival rate than the group with low SUV39H1 expression in terms of 10-year disease-free survival (DFS) according to the database analyses (P=0.035). However, SUV39H1 was revealed as an independent predictor of overall survival (OS) and progression-free survival (PFS) in patients from both databases and our hospital (P>0.05).Conclusion: SUV39H1 expression is higher in DLBCL tissues than in normal and other cancer tissues, indicating that DLBCL patients may not develop bulky tumors over follow-up and suggesting that SUV39H1 might serve not only as a predictive factor in the clinic but also as a target for the epigenetic therapy of DLBCL.

Reinforcement learning has achieved remarkable performance in a wide range of tasks these days. Nevertheless, some unsolved problems limit its applications in real-world control. One of them is model misspecification, a situation where an agent is trained and deployed in environments with different transition dynamics. We propose an novel framework that utilize history trajectory and Partial Observable Markov Decision Process Modeling to deal with this dilemma. Additionally, we put forward an efficient adversarial attack method to assist robust training. Our experiments in four gym domains validate the effectiveness of our framework.

Controllable synthesis of FeS nanoparticles with high electrochemical performance is still a challenge for cheap Li-ion batteries. In this paper, FeS nanospheres/Fe/hard carbon mesoporous sheet (FeSNs-Fe/HCMS) nanocomposites are synthesized by using sulfate pulping red liquor (SPRL) as a multifunctional template. In the nanocomposite, FeS nanospheres (20–50 nm) are homogeneously embedded in Fe/hard carbon mesoporous sheets. The heterogeneous nucleation and confined growth mechanisms of FeS nanospheres are established. The synthesis mechanism and electrochemical reactions of FeSNs-Fe/HCMS nanocomposites are proposed. The influences of the additive amount of sulfate pulping red liquor on the structure and electrochemical performance of FeSNs-Fe/HCMS nanocomposites are investigated. The results of electrochemical investigation show that this nanocomposite anode exhibits a high initial discharge capacity of 1182 mAh g−1 at a current density of 1 A g−1, which is far beyond the theoretical specific capacity of FeS (609 mAh g−1). Even after cycling at 1 A g−1 over 100 cycles, this anode still retains a high discharge specific capacity of 881 mAh g−1. This work not only reduces pollution of sulfate pulping red liquor, but also benefits the utilization of waste, which proves that sulfate pulping red liquor has good application prospects in Li-ion batteries.

Qianwen Ma,1,2,* Yu Chang,1,2,* Ling Li,1,2 Xin Li,1,2 Xinhua Wang,1,2 Jingjing Wu,1,2 Xiaorui Fu,1,2 Zhenchang Sun,1,2 Hui Yu,1,2 Xudong Zhang,1,2 Zhiyuan Zhou,1,2 Feifei Nan,1,2 Zhaoming Li,1,2 Xiyang Liu,1,2 Qian Zhao,1,2 Yang Li,1,2 Lan Zhang,2,3 Mingzhi Zhang,1,2 Lei Zhang1,2 1Department of Oncology, The First Affiliated Hospital of Zhengzhou University, Zhengzhou 450000, Henan, China; 2Lymphoma Diagnosis and Treatment Center of Henan Province, Zhengzhou 450000, Henan, China; 3Department of Pathology, The First Affiliated Hospital of Zhengzhou University, Zhengzhou 450000, Henan, China *These authors contributed equally to this work Purpose: To compare the efficacy of rituximab, dose-adjusted etoposide, prednisone, vincristine, cyclophosphamide, and doxorubicin (R-DA-EPOCH) with traditional rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone (R-CHOP) regimens in double-hit lymphoma (DHL) and gene copy number gain (CNG) lymphoma and to contrast the prognosis of these two disease types. Methods: We retrospectively examined 127 cases of newly diagnosed diffuse large B-cell lymphoma (DLBCL), and used fluorescence in situ hybridization (FISH) to detect genetic abnormalities in MYC, BCL2, and BCL6. Results: In the two schemes, the 2-year progression-free survival (PFS) was higher for R-DA-EPOCH group than for R-CHOP (79.8% vs 57.5%, P=0.002), this advantage was also reflected in 2-year overall survival (OS) (81.6% vs 58.5%, P=0.002). In double CNG patients, R-DA-EPOCH regimen was significantly better than R-CHOP (P=0.007 for PFS, P=0.010 for OS), and R-DA-EPOCH has the same advantage in DHL patients (P=0.001 for PFS, P=0.047 for OS). For the two disease types, the PFS for DHL was inferior to that for double CNG (52.9% vs 72.4%, P=0.008), while the OS was not significantly different (P=0.050). Subgroup analysis showed that the PFS for double CNG with MYC and BCL2 was superior to that for DHL with MYC and BCL2 (P=0.043), this trend is also seen in double CNG and DHL with MYC and BCL6 (P=0.036). However, the OS was not significantly different between the two subgroups. Multivariate analyses showed that in DLBCL patients with genetic abnormality detected by FISH, the treatment and disease types were independent prognostic factors. The adverse reaction rates were similar in R-DA-EPOCH and R-CHOP (P>0.05). Conclusion: Our retrospective study shows that DHL has a poorer prognosis than double CNG. Based on its improved lifetime and good tolerance, R-DA-EPOCH is a promising regimen for DHL or double CNG, which is expected to become the first-line treatment for high-risk DLBCL types based on more clinical research. Keywords: R-DA-EPOCH, R-CHOP, DHL, double CNG, efficacy and prognosis

Smart manufacturing is the core idea of the fourth industrial evolution. For a smart manufacturing shop floor, real-time monitoring, simulation and prediction of manufacturing operations are vital ...

As the manufacturing industry is heading toward the fourth industrial revolution, smart manufacturing is born at the right moment. By integrating new information technologies, such as the Internet of Things, the cyber-physical system (CPS), big data, and artificial intelligence, smart manufacturing has endowed the factories and shop floors with much intelligence, which is characterized by the organic cooperation among workers, machines, unmanufactured products, and other physical assets. In this situation, endowing these smart physical assets with self-X intelligence and autonomy to make manufacturing resources allocation decisions autonomously has been a vital problem that needs prompt solutions. To solve this problem, this paper deals with training a reasonable knowledge model from the historical shop floor data using a hidden Markov model (HMM) theory. In this model, the unmanufactured product's machining feature/process flow is considered as an observation sequence and the corresponding smart manufacturing resources (SMRs) sequence is considered as a hidden state sequence. The solving method to train the HMM-based knowledge model for autonomous manufacturing resources allocation (A-MRA) is further described in a step-by-step manner. Thereafter, a demonstrative case is studied to verify the proposed model and method. First, 123 pairs of historical data (i.e., process flow and SMR sequence) are used to learn the HMM-based knowledge model and another 5 pairs of historical data are used to test the feasibility and accuracy of the proposed model. The results show that only three elements (total $5\times 9$ elements) in the predicted SMR sequences are different from those in the historical SMR sequences, and the average vector angle between the five predicted and historical SMR sequences is 11.68°, which is relatively low considering that only nine elements exists in each SMR sequence.

Extra-hypothalamic growth hormone-releasing hormone (GHRH) plays an important role in reproduction. To study the treatment effect of Grin (a novel hGHRH homodimer), the infertility models of 85 male Chinese hamsters were established by intraperitoneally injecting 20 mg/kg of cyclophosphamide once in a week for 5 weeks and the treatment with Grin or human menopausal gonadotropin (hMG) as positive control was evaluated by performing a 3-week mating experiment. 2-8 mg/kg of Grin and 200 U/kg of hMG showed similar effect and different pathological characteristics. Compared to the single cyclophosphamide group (0%), the pregnancy rates (H-, M-, L-Grin 26.7, 30.8, 31.3%, and hMG 31.3%) showed significant difference, but there was no difference between the hMG and Grin groups. The single cyclophosphamide group presented loose tubules with pathologic vacuoles and significant TUNEL positive cells. Grin induced less weight of body or testis, compactly aligned tubules with little intra-lumens, whereas hMG caused more weight of body or testis, enlarging tubules with annular clearance. Grin presented a dose-dependent manner or cell differentiation-dependentincrease in testicular GHRH receptor, and did not impact the levels of blood and testicular GH, testosterone. Grin promotes fertility by proliferating and differentiating primitive cells through up-regulating testicular GHRH receptor without triggering GH secretion, which might solve the etiology of oligoasthenozoospermia.

Chloride (Cl−) is required for photosynthesis and regulates osmotic balance. However, excess Cl− application negatively interacts with nitrate (NO−3) uptake, although its effect on NO−3 metabolism remains unclear. The aim was to test whether Cl− stress disturbs nitrate reductase activity (NRA). A maize variety (Zea mays L. cv. LG 30215) was hydroponically cultured in a greenhouse under the following conditions: control (2 mM CaCl2), moderate Cl− (10 mM CaCl2), high Cl− (60 mM CaCl2). To substantiate the effect of Cl− stress further, an osmotic stress with lower intensity was induced by 60 g polyethylene glycol (PEG) 6000 L−1 + 2 mM CaCl2), which was 57% of the osmotic pressure being produced by 60 mM CaCl2. Results show that high Cl− and PEG-induced osmotic stress significantly reduced shoot biomass, stomatal conductance and transpiration rate, but NRA was only decreased by high Cl− stress. The interference of NRA in chloride-stressed maize is supposed to be primarily caused by the antagonistic uptake of Cl− and NO−3.

Off-policy evaluation (OPE) is to evaluate a target policy with data generated by other policies. Most previous OPE methods focus on precisely estimating the true performance of a policy. We observe that in many applications, (1) the end goal of OPE is to compare two or multiple candidate policies and choose a good one, which is a much simpler task than precisely evaluating their true performance; and (2) there are usually multiple policies that have been deployed to serve users in real-world systems and thus the true performance of these policies can be known. Inspired by the two observations, in this work, we study a new problem, supervised off-policy ranking (SOPR), which aims to rank a set of target policies based on supervised learning by leveraging off-policy data and policies with known performance. We propose a method to solve SOPR, which learns a policy scoring model by minimizing a ranking loss of the training policies rather than estimating the precise policy performance. The scoring model in our method, a hierarchical Transformer based model, maps a set of state-action pairs to a score, where the state of each pair comes from the off-policy data and the action is taken by a target policy on the state in an offline manner. Extensive experiments on public datasets show that our method outperforms baseline methods in terms of rank correlation, regret value, and stability. Our code is publicly available at GitHub., Comment: Accepted by ICML 2022