Search

Direct oral anticoagulants (DOACs) could effectively prevent the occurrence of cancer-associated venous thromboembolism (CAVTE), which incidence rate was estimated to be 4-20%. But the efficacy and safety remain controversial between DOACs and low molecular weight heparin (LMWH).PubMed, Cochrane Library, Embase, ClinicalTrials.gov databases for randomized controlled trials (RCTs) were systematically searched from inception to March 15, 2022. A random-effects model was used to report the odds ratio (OR) and 95% confidence interval (CI) for both direct and network meta-analyses.Seven studies were included totaling 3242 patients. A lower rate of recurrence VTE was noted in the DOACs compared with LMWH (OR 0.62, 95% CI 0.47-0.82, ICompared with LMWH, DOACs seemed to have a decreased risk of recurrence VTE while increasing CRNMB and CRB. DOACs and LMWH were equivalent to the aspect of MB, but DOACs had a higher MB risk in patients with gastrointestinal and genitourinary malignancies. Apixaban may be the lowest risk compared to the other DOACs in precaution of VTE and reducing bleeding events.

The fracture risks of non-vitamin K antagonist oral anticoagulants (NOACs) vs. warfarin in patients with atrial fibrillation (AF) remain controversial. PubMed, Cochrane Library, EMBASE, Clinical Trials.gov databases for RCTs, and cohort studies were systematically searched from inception to 10 June 2021. Twelve-two studies met the inclusion criteria and 477,821 patients were included. Warfarin increased the risk of fracture in AF patients compared with NOACs in overall any fracture (RR = 0.79; 95% CI = 0.70–10.88; p = 0.00), osteoporotic fracture (RR = 0.746; 95% CI = 0.630–0.883; p = 0.001). No significant difference was observed in the hip or pelvic fracture, vertebral fracture, extremity fracture, wrist fracture, femoral neck fracture, and ankle fracture. In subgroup analyses based on several aspects, NOACs were associated with a significant reduction in any fracture (standard dosage NOACs, cohort studies, elderly patients, rivaroxaban in RCTs, dabigatran, rivaroxaban, and apixaban in cohort studies), in the hip/pelvic fracture (follow-up time ≤1 year, cohort studies), and osteoporotic fracture (cohort studies). NOACs were associated with a significantly lower risk of any fracture and osteoporotic fracture compared to warfarin. This benefit was also observed in specific NOACs types of dabigatran, rivaroxaban, and apixaban. However, whether NOACs had a less fracture risk than warfarin on the other risk of fractures was still uncertain.

Recently, privacy issues in web services that rely on users' personal data have raised great attention. Unlike existing privacy-preserving technologies such as federated learning and differential privacy, we explore another way to mitigate users' privacy concerns, giving them control over their own data. For this goal, we propose a privacy aware recommendation framework that gives users delicate control over their personal data, including implicit behaviors, e.g., clicks and watches. In this new framework, users can proactively control which data to disclose based on the trade-off between anticipated privacy risks and potential utilities. Then we study users' privacy decision making under different data disclosure mechanisms and recommendation models, and how their data disclosure decisions affect the recommender system's performance. To avoid the high cost of real-world experiments, we apply simulations to study the effects of our proposed framework. Specifically, we propose a reinforcement learning algorithm to simulate users' decisions (with various sensitivities) under three proposed platform mechanisms on two datasets with three representative recommendation models. The simulation results show that the platform mechanisms with finer split granularity and more unrestrained disclosure strategy can bring better results for both end users and platforms than the "all or nothing" binary mechanism adopted by most real-world applications. It also shows that our proposed framework can effectively protect users' privacy since they can obtain comparable or even better results with much less disclosed data., Accepted by TOIS 2022

Reinforcement learning (RL) has recently been introduced to interactive recommender systems (IRS) because of its nature of learning from dynamic interactions and planning for long-run performance. As IRS is always with thousands of items to recommend (i.e., thousands of actions), most existing RL-based methods, however, fail to handle such a large discrete action space problem and thus become inefficient. The existing work that tries to deal with the large discrete action space problem by utilizing the deep deterministic policy gradient framework suffers from the inconsistency between the continuous action representation (the output of the actor network) and the real discrete action. To avoid such inconsistency and achieve high efficiency and recommendation effectiveness, in this paper, we propose a Tree-structured Policy Gradient Recommendation (TPGR) framework, where a balanced hierarchical clustering tree is built over the items and picking an item is formulated as seeking a path from the root to a certain leaf of the tree. Extensive experiments on carefully-designed environments based on two real-world datasets demonstrate that our model provides superior recommendation performance and significant efficiency improvement over state-of-the-art methods.

Intra-camera supervision (ICS) for person re-identification (Re-ID) assumes that identity labels are independently annotated within each camera view and no inter-camera identity association is labeled. It is a new setting proposed recently to reduce the burden of annotation while expect to maintain desirable Re-ID performance. However, the lack of inter-camera labels makes the ICS Re-ID problem much more challenging than the fully supervised counterpart. By investigating the characteristics of ICS, this paper proposes camera-specific non-parametric classifiers, together with a hybrid mining quintuplet loss, to perform intra-camera learning. Then, an inter-camera learning module consisting of a graph-based ID association step and a Re-ID model updating step is conducted. Extensive experiments on three large-scale Re-ID datasets show that our approach outperforms all existing ICS works by a great margin. Our approach performs even comparable to state-of-the-art fully supervised methods in two of the datasets., Comment: Accepted by WACV2021

Interactive recommender system (IRS) has drawn huge attention because of its flexible recommendation strategy and the consideration of optimal long-term user experiences. To deal with the dynamic user preference and optimize accumulative utilities, researchers have introduced reinforcement learning (RL) into IRS. However, RL methods share a common issue of sample efficiency, i.e., huge amount of interaction data is required to train an effective recommendation policy, which is caused by the sparse user responses and the large action space consisting of a large number of candidate items. Moreover, it is infeasible to collect much data with explorative policies in online environments, which will probably harm user experience. In this work, we investigate the potential of leveraging knowledge graph (KG) in dealing with these issues of RL methods for IRS, which provides rich side information for recommendation decision making. Instead of learning RL policies from scratch, we make use of the prior knowledge of the item correlation learned from KG to (i) guide the candidate selection for better candidate item retrieval, (ii) enrich the representation of items and user states, and (iii) propagate user preferences among the correlated items over KG to deal with the sparsity of user feedback. Comprehensive experiments have been conducted on two real-world datasets, which demonstrate the superiority of our approach with significant improvements against state-of-the-arts.

Accurately characterizing the user's current interest is the core of recommender systems. However, users' interests are dynamic and affected by intent factors and preference factors. The intent factors imply users' current needs and change among different visits. The preference factors are relatively stable and learned continuously over time. Existing works either resort to the sequential recommendation to model the current browsing intent and historical preference separately or just mix up these two factors during online learning. In this paper, we propose a novel learning strategy named FLIP to decouple the learning of intent and preference under the online settings. The learning of the intent is considered as a meta-learning task and fast adaptive to the current browsing; the learning of the preference is based on the calibrated user intent and constantly updated over time. We conducted experiments on two public datasets and a real-world recommender system. When equipping it with modern recommendation methods, significant improvements are demonstrated over strong baselines.

The vacuum preloading method combining prefabricated vertical drains (PVDs) and surcharge preloading from vacuum pressure is widely used in the improvement of dredger fill foundation. However, a clogging effect arises within the soils around the PVDs during the vacuum preloading, which significantly affects the soil consolidation. Previous studies indicated that the organic matter and soil pH circumstance influence the soil-water interaction and soil microstructure. In order to investigate the influences of organic matter and soil pH on the consolidation of clayey dredger fill, a series of small scale vacuum preloading tests was conducted on soils with different organic contents and soil pH values. Subsequently, Scanning Electron Microscopy observation test and vane shear test were carried out to evaluate the effect of vacuum preloading consolidation. The results showed that a low organic content or a high soil pH is beneficial for soil consolidation by vacuum preloading method, resulting in a more compact microstructure and a greater vane strength. There are two actions on the organic matter during the vacuum preloading consolidation, namely, material migration and organic matter dissolution. The organic matter dissolution effect is predominate at soil pH > 9.1, so that the consolidation effect is improved. However, the material migration effect is predominate at pH = 8.1, clay particles and organic matter can be transported with the water flow and redeposit near the drain pipes, resulting in a clogging effect. Moreover, a higher pH can induce the change of microstructure from flocculated association to deflocculated and dispersed association of clay particles, which is positive to soil consolidation. The results in this paper may offer helpful references for engineering practice.

Water molecules are adsorbed onto the surface and interlayer of clay minerals and form a film, which causes that pore pressure diffusion in clayey soils differs from that in granular material. In this paper, adsorbed water is classified into strongly adsorbed water and loosely adsorbed water, which provide a better understanding of the physical mechanisms of the adsorbed water. To investigate the impact of adsorbed water on pore pressure diffusion, a series laboratory tests were conducted, including the uplift pressure test, pore pressure diffusion test and adsorbed water content test. The experimental results implies that strongly adsorbed water can neither flow nor participate in pressure diffusion, but loosely adsorbed water shears at pressure gradients and participates in laminar flow. Therefore, as long as the dense clayey soil contains loosely adsorbed water, the pore pressure change at a local site can diffuse throughout the material. A concept called “loosely adsorbed water index (I lo) is defined to characterize both the content and physical properties of loosely adsorbed water: (1) when I lo ≥ 1, the peripheral water molecules of loosely adsorbed water layer essentially behave the same as free water; (2) as I lo → 0, the physical properties of adsorbed water may evolve from a gel-like state to a solid-like state (strongly adsorbed water). According to the density of clay and the content of adsorbed water, the process of pore pressure diffusion can be divided into three modes: flow channels of free fluid, deformations of the adsorbed water layer and shear motions of the peripheral-adsorbed water molecules. Pore pressure response to additional water pressure strongly depends on the permeability, seepage path and boundary conditions. The time-lag effect of pore pressure diffusion should be considered in applications of effective stress in engineering.

Reinforcement learning techniques have recently been introduced to interactive recommender systems to capture the dynamic patterns of user behavior during the interaction with recommender systems and perform planning to optimize long-term performance. Most existing research work focuses on designing policy and learning algorithms of the recommender agent but seldom cares about the state representation of the environment, which is indeed essential for the recommendation decision making. In this paper, we first formulate the interactive recommender system problem with a deep reinforcement learning recommendation framework. Within this framework, we then carefully design four state representation schemes for learning the recommendation policy. Inspired by recent advances in feature interaction modeling in user response prediction, we discover that explicitly modeling user–item interactions in state representation can largely help the recommendation policy perform effective reinforcement learning. Extensive experiments on four real-world datasets are conducted under both the offline and simulated online evaluation settings. The experimental results demonstrate the proposed state representation schemes lead to better performance over the state-of-the-art methods.

In this paper, we propose a hybrid framework for query processing and data analytics over large-scale data on Spark, to support multi-paradigm process (incl. SQL, OLAP, data mining, machine learning etc.) in distributed environments. The framework features a three-layer data process module and a work flow module which controls the former. We will demonstrate the strength of our framework properly applying traffic scenarios in a real world.