A5100735406- Recommender Systems and Techniques
- Advanced Bandit Algorithms Research
- Advanced Neural Network Applications
- Advanced Decision-Making Techniques
- Digital Holography and Microscopy
- Advanced Image and Video Retrieval Techniques
- Advanced Measurement and Detection Methods
- Power Systems and Technologies
- Generative Adversarial Networks and Image Synthesis
- Digital Imaging for Blood Diseases
- Smart Grid and Power Systems
- Data Management and Algorithms
- Image Retrieval and Classification Techniques
- Music and Audio Processing
- Domain Adaptation and Few-Shot Learning
- Plant nutrient uptake and metabolism
- Human Mobility and Location-Based Analysis
- Evaluation Methods in Various Fields
- Stochastic Gradient Optimization Techniques
- Plant Molecular Biology Research
- Higher Education and Teaching Methods
- Image and Object Detection Techniques
- Topic Modeling
- Advanced Graph Neural Networks
- Image Processing Techniques and Applications
Tsinghua University
2017-2024
State Key Laboratory of Plant Physiology and Biochemistry
2019
Zhejiang University
2019
Harbin Engineering University
2019
Sichuan University
2012
Hefei University of Technology
2011
Ministry of Education of the People's Republic of China
2010
Most modern recommender systems predict users preferences with two components: user and item embedding learning, followed by the user-item interaction modeling. By utilizing auxiliary review information accompanied ratings, many of existing review-based recommendation models enriched user/item learning ability historical reviews or better modeled interactions help available target reviews. Though significant progress has been made, we argue that current solutions for suffer from drawbacks....
The major root system of cereals consists crown roots (or adventitious roots), which are important for anchoring plants in the soil and water nutrient uptake. However, molecular basis formation is largely unknown. Here, we isolated a rice (
In modern recommender systems, sequential recommendation leverages chronological user behaviors to make effective next-item suggestions, which suffers from data sparsity issues, especially for new users. One promising line of work is the cross-domain recommendation, trains models with across multiple domains improve performance in data-scarce domains. Recent proposed such as PiNet and DASL have a common drawback relying heavily on overlapped users different domains, limits their usage...
Pick-up and delivery (PD) services such as online food ordering are playing an increasingly important role in serving people's daily demands. Accurate PD route prediction (PDRP) is for service providers to efficiently schedule riders improve quality. It crucial model the decision-making process behind choice of PDRP. Recent years have witnessed success utilizing imitation learning (IL) user process. Therefore, we propose deploy IL framework solve PDRP problem. However, there still exist...
It is difficult to search for interference fringes in Linnik white light interferometry with an extremely short coherence length because of the optical path mismatch two arms and defocus reference mirror test surface. We present automated method tackle this problem paper. The determination best foci surface implemented by astigmatic based on a modified commercial DVD pickup head embedded system. improved setting threshold value sum signal truncate normalized focus error (NFES). truncated...
Modern personalized recommendation services often rely on user feedback, either explicit or implicit, to improve the quality of services. Explicit feedback refers behaviors like ratings, while implicit clicks. However, in scenario full-screen video viewing experiences Tiktok and Reels, click action is absent, resulting unclear from users, hence introducing noises modeling training. Existing approaches de-noising mainly focus positive instances ignoring noise a large amount sampled negative...
Deep neural networks are widely used in personalized recommendation systems. Unlike regular DNN inference workloads, is memory-bound due to the many random memory accesses needed lookup embedding tables. The also heavily constrained terms of latency because producing a for user must be done about tens milliseconds. In this paper, we propose MicroRec, high-performance engine MicroRec accelerates by (1) redesigning data structures involved embeddings reduce number lookups and (2) taking...
Detailed damage analyses of an Y 2 O 3 stabilized ZrO top coat (TC)–MCrAlY bond (BC)–superalloy thermal barrier coating (TBC) system during thermomechanical fatigue (TMF) and gradient mechanical (TGMF) tests had been performed in present work. During tests, the lifetime TBCs was strongly dependent on strain ranges, pre-oxidation time TBCs. Cracks were initiated TGO layer, propagated along TC/TGO or TGO/BC interface, forming delamination cracks. When cracks connected with segmentation which...
During the construction of computer network quality course to improve course, paper designs and develops website. The development website is based on B/S architecture MVC design pattern adopts Struts framework. enough incarnate user-centered idea provides an excellent intercourse platform for teachers students. shows function modules, specifically describes main modules update maintenance
In collaborative filtering (CF), interaction function (IFC) plays the important role of capturing interactions among items and users. The most popular IFC is inner product, which has been successfully used in low-rank matrix factorization. However, real-world applications can be highly complex. Thus, other operations (such as plus concatenation), may potentially offer better performance, have proposed. Nevertheless, it still hard for existing IFCs to consistently good performance across...
In order to obtain accurate measurement results of particle size, the watershed algorithm is used segment image after steps including median filtering, binaryzation, expansion and corrosion, distance transformation seed point marking. Then segmented data filtered errors with large deviations are eliminated. The result filtering compared without filtering. verified. Finally, precise size obtained. experiments show proposed method can accurately measure overlapped particles.
Attention Mechanism is a widely used method for improving the performance of convolutional neural networks (CNNs) on computer vision tasks. Despite its pervasiveness, we have poor understanding what effectiveness stems from. It popularly believed that from visual attention explanation, advocating focusing important part input data rather than ingesting entire input. In this paper, find there only weak consistency between weights features and their importance. Instead, verify crucial role...