A5101904758- Recommender Systems and Techniques
- Advanced Graph Neural Networks
- Advanced Bandit Algorithms Research
- Context-Aware Activity Recognition Systems
- Expert finding and Q&A systems
- Cerebrovascular and Carotid Artery Diseases
- Seismology and Earthquake Studies
- Caching and Content Delivery
- Methane Hydrates and Related Phenomena
- Acute Ischemic Stroke Management
- Neural Networks and Reservoir Computing
- Respiratory Support and Mechanisms
- Interstitial Lung Diseases and Idiopathic Pulmonary Fibrosis
- Nonlinear Dynamics and Pattern Formation
- Esophageal and GI Pathology
- Pancreatic and Hepatic Oncology Research
- Neonatal Respiratory Health Research
- High-Temperature Coating Behaviors
- Topic Modeling
- Metal and Thin Film Mechanics
- Chronic Obstructive Pulmonary Disease (COPD) Research
- Crystallization and Solubility Studies
- X-ray Diffraction in Crystallography
- Image and Video Quality Assessment
- Impact of Light on Environment and Health
Heidelberg University
2024-2025
University Hospital Ulm
2024-2025
University Hospital Heidelberg
2025
École Polytechnique Fédérale de Lausanne
2022-2025
Second Hospital of Hebei Medical University
2023-2024
Hebei Medical University
2023-2024
Kuaishou (China)
2021-2024
North China Electric Power University
2024
Shanxi Academy of Medical Sciences
2023-2024
Shanxi Medical University
2023-2024
Sequential recommendation aims to leverage users' historical behaviors predict their next interaction. Existing works have not yet addressed two main challenges in sequential recommendation. First, user rich sequences are often implicit and noisy preference signals, they cannot sufficiently reflect actual preferences. In addition, dynamic preferences change rapidly over time, hence it is difficult capture patterns sequences. this work, we propose a graph neural network model called SURGE...
Sequential recommendation aims at identifying the next item that is preferred by a user based on their behavioral history. Compared to conventional sequential models leverage attention mechanisms and RNNs, recent efforts mainly follow two directions for improvement: multi-interest learning graph convolutional aggregation. Specifically, methods such as ComiRec MIMN, focus extracting different interests performing historical clustering, while convolution including TGSRec SURGE elect refine...
Pancreatic ductal adenocarcinoma (PDAC) is highly aggressive and characterized by pronounced desmoplasia. PDAC cells communicate with cancer-associated fibroblasts (CAFs) in a paracrine/reciprocal manner, substantially promoting tumor growth desmoplastic responses. This study highlights the critical role of anterior gradient 2 (AGR2), an endoplasmic reticulum protein disulfide isomerase, secreted to activate CAFs via Wnt signaling pathway. Activated CAFs, turn, secrete insulin-like factor 1...
Recently, micro-video sharing platforms such as Kuaishou and Tiktok have become a major source of information for people's lives. Thanks to the large traffic volume, short video lifespan streaming fashion these services, it has more pressing improve existing recommender systems accommodate challenges in cost-effective way. In this paper, we propose novel concept-aware denoising graph neural network (named Conde) recommendation. Conde consists three-phase convolution process derive user...
Sequential recommendation is one of the most important tasks in recommender systems, which aims to recommend next interacted item with historical behaviors as input. Traditional sequential always mainly considers collected positive feedback such click, purchase, etc. However, short-video platforms TikTok, video viewing behavior may not represent feedback. Specifically, videos are played automatically, and users passively receive recommended videos. In this new scenario, express negative by...
The confluence of Search and Recommendation (S&R) services is vital to online services, including e-commerce video platforms. integration S&R modeling a highly intuitive approach adopted by industry practitioners. However, there noticeable lack research conducted in this area within academia, primarily due the absence publicly available datasets. Consequently, substantial gap has emerged between academia regarding endeavors joint optimization using user behavior data from both services. To...
Short-video recommendation is one of the most important applications in today's industrial information systems. Compared with other tasks, enormous amount feedback typical characteristic. Specifically, short-video recommendation, easiest-to-collect user theskipping behavior, which leads to two critical challenges for model. First, skipping behavior reflects implicit preferences, and thus, it challenging interest extraction. Second, this kind special involves multiple objectives, such as...
A simple design of a hybrid integrated short-external-cavity chaotic semiconductor laser module is presented and fabricated. distributed feedback chip directed to the transflective mirror through collimating micro-lens, part light reflected by into induce chaos. Transmitted mirror, generated coupled pigtail fiber as output focusing micro-lens. It should be pointed out that between emitting surface works short external straight cavity. All components above together with submount, heat sink,...
The emergence of the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) and its subsequent Omicron variant has raised concerns for chronic obstructive pulmonary disease (COPD) patients due to potential risk disruptions healthcare services unknown comorbidities between COPD Omicron.
Abstract The ARP2/3 complex, which orchestrates actin cytoskeleton organization and lamellipodia formation, has been implicated in the initiation of pancreatic ductal adenocarcinoma (PDAC). This study aims to clarify its impact on activity cancer‐associated fibroblasts (CAFs), key players PDAC progression, patient outcomes. Early carcinogenesis was modeled p48 Cre ; LSL‐Kras G12D mice with caerulein‐induced pancreatitis, complemented by vitro studies human immortalized stellate cells (PSCs)...
The structure, electronic and optical properties of 3C–SiC nano films were calculated according to first principles based on density functional theory. It is manifested that the structurally stable when number layers was increased 12. Compared with bulk counterparts, displayed narrower band gaps, their Fermi energy levels brought into conduction band. ε2(ω) dielectric function, extinction coefficient, absorption coefficient greater than in ultraviolet range. With layer numbers decreased,...
Defect detection is a common task in industry production, array defect an undesirable phenomenon substrate production process due to the environment, conditions and so on. Array very important for quality performance of final product. Compared with other defects, defects have more complex background, make logic advanced difficult judge. In this paper, we propose convolution neural network (CNN) based on Faster‐RCNN architecture. On basis VGG16 network, add cross‐connection layer feature...
Insulin resistance (IR) is associated with atherosclerotic plaque progression and the occurrence of stroke, triglyceride-glucose (TyG) index serving as a surrogate indicator. The present study aimed to investigate association between TyG levels intracranial arterial remodeling in patients acute ischemic stroke (AIS).Patients AIS who visited Neurology Department Second Hospital Hebei Medical University underwent high-resolution magnetic resonance imaging (HR-MRI) September 2018 October 2021...
Nowadays, many platforms provide users with both search and recommendation services as important tools for accessing information. The phenomenon has led to a correlation between user behaviors, providing an opportunity model interests in fine-grained way. Existing approaches either behaviors separately or overlook the different transitions behaviors. In this paper, we propose framework named UniSAR that effectively models types of behavior Unified Search And Recommendation service....
Background: Both intracranial atherosclerosis and white matter hyperintensity (WMH) are prevalent among the stroke population. However, relationship between WMH has not been fully elucidated. Therefore, aim of this study was to investigate characteristics atherosclerotic plaques severity in patients with ischemic using high-resolution magnetic resonance vessel wall imaging. Methods: Patients hospitalized concurrent at Beijing Tsinghua Changgung Hospital, a tertiary comprehensive center, who...
Industrial recommendation systems (RS) rely on the multi-stage pipeline to balance effectiveness and efficiency when delivering items from a vast corpus users. Existing RS benchmark datasets primarily focus exposure space, where novel algorithms are trained evaluated. However, these transition real world industrial RS, they face critical challenge of handling unexposed which significantly larger space than exposed one. This discrepancy profoundly impacts their practical performance....