A5032497440- Advanced Sensor and Control Systems
- Advanced Algorithms and Applications
- Military Defense Systems Analysis
- Advanced Measurement and Detection Methods
- Recommender Systems and Techniques
- Topic Modeling
- Guidance and Control Systems
- Advanced Computational Techniques and Applications
- Image and Object Detection Techniques
- Advanced Graph Neural Networks
- Advanced Decision-Making Techniques
- Target Tracking and Data Fusion in Sensor Networks
- Military Strategy and Technology
- Embedded Systems and FPGA Design
- Robotic Path Planning Algorithms
- Evaluation Methods in Various Fields
- Power Systems and Technologies
- Distributed and Parallel Computing Systems
- Infrared Target Detection Methodologies
- Sensor Technology and Measurement Systems
- Industrial Technology and Control Systems
- Inertial Sensor and Navigation
- Automated Road and Building Extraction
- Intelligent Tutoring Systems and Adaptive Learning
- Optical Systems and Laser Technology
Shanghai Jiao Tong University
2022-2024
The University of Sydney
2024
North China University of Technology
2023
Hefei University of Technology
2023
Xi’an Jiaotong-Liverpool University
2023
Xidian University
2022
China Agricultural University
2021
Naval University of Engineering
2007-2018
Beijing Institute of Technology
2018
Institute of Theoretical Physics
2018
Click-through rate (CTR) prediction has become increasingly indispensable for various Internet applications. Traditional CTR models convert the multi-field categorical data into ID features via one-hot encoding, and extract collaborative signals among features. Such a paradigm suffers from problem of semantic information loss. Another line research explores potential pretrained language (PLMs) by converting input textual sentences through hard prompt templates. Although are preserved, they...
With the rapid development of remote sensing technologies, SAR satellites like China’s Gaofen-3 satellite have more imaging modes and higher resolution. availability high-resolution images, automatic ship target detection has become an important topic in maritime research. In this paper, a novel method based on gradient integral features is proposed. This mainly composed three steps. First, preprocessing step, filter employed to smooth clutters smoothing effect can be adaptive adjusted...
An aanytime algorithm based on modified genetic (GA) for dynamic WTA problem, subject to temporal constraints, is developed in the paper. In weapons are assigned targets one by before deadline of each target comes. After a with some weapon, replaced new all chromosomes population while optimization process will not undergo any restart. The has three main advantages: 1) can be dynamically accommodated allocation without losing previous optimizing information, 2) quality pairing decisions may...
Click-through rate (CTR) prediction plays as a core function module in various personalized online services. The traditional ID-based models for CTR take inputs the one-hot encoded ID features of tabular modality, which capture collaborative signals via feature interaction modeling. But encoding discards semantic information included textual features. Recently, emergence Pretrained Language Models (PLMs) has given rise to another paradigm, takes sentences modality obtained by hard prompt...
Computerized Adaptive Testing (CAT) refers to an online system that adaptively selects the best-suited question for students with various abilities based on their historical response records. Compared traditional CAT methods heuristic rules, recent data-driven obtain higher performance by learning from large-scale datasets. However, most only focus quality objective of predicting student ability accurately, but neglect concept diversity or exposure control, which are important considerations...
An airport is a typical strategic target for detection and attack, runway recognition automatically in an aerial image has important military meaning. Aimed at the characteristics of edge lines image, algorithm based on heuristic line extraction presented. First, we use Sobel operator to get that been number normalized, extract available straight by employing search image. Then, connect parallel lines. Finally, recognize correctly grey area. Many experiments show this method can find exactly quickly
With the success of deep learning, click-through rate (CTR) predictions are transitioning from shallow approaches to architectures. Current CTR prediction usually follows Embedding & MLP paradigm, where model embeds categorical features into latent semantic space. This paper introduces a novel embedding technique called neural statistics that instead learns explicit semantics by incorporating feature engineering as an innate prior architecture in end-to-end manner. Besides, since statistical...
The Cascade H-bridge converter (CHB) is attractive for photovoltaic applications because of the advantages its high modularity and high-voltage ability. There exists a large number capacitors in CHB system, which determines output waveform quality. Therefore, to get health status those quite meaningful. In this paper, capacitor condition monitoring method based on recursive least square (RLS) method. equation HB module capacitance value proposed, utilizes feature that ac side inverter only...
In order to solve nonlinear integer programming which is widely used in engineering fields, a particle swarm optimization based on genetic operators put forward. The restriction of problems transformed by coding solutions. According optimal solutions population and individual, the new updated crossover, mutation selection operators. After numerical experiment algorithm, it proved be feasible effective, especially solving large-scale problems, shows much better performance.
A Hopfield neural network-based approach for task scheduling in a Grid environment is proposed this paper. All constraints and the optimization object of problem are developed included computational energy function network. To avoid network converge into local minimum volume, simulated annealing algorithms applied to Thus global as feasible solution achieved. The theoretic analyses simulation experiments have manifested approach's effectiveness.
The significant advancements in large language models (LLMs) give rise to a promising research direction, i.e., leveraging LLMs as recommenders (LLMRec). efficacy of LLMRec arises from the open-world knowledge and reasoning capabilities inherent LLMs. acquires recommendation through instruction tuning based on user interaction data. However, order protect privacy optimize utility, it is also crucial for intentionally forget specific data, which generally referred unlearning. In era LLMs,...
Recommender systems play important roles in various applications such as e-commerce, social media, etc. Conventional recommendation methods usually model the collaborative signals within tabular representation space. Despite personalization modeling and efficiency, latent semantic dependencies are omitted. Methods that introduce semantics into then emerge, injecting knowledge from space where general language understanding compressed. However, existing semantic-enhanced focus on aligning two...
Recommender systems play important roles in various applications such as e-commerce, social media, etc. Conventional recommendation methods usually model the collaborative signals within tabular representation space. Despite personalization modeling and efficiency, latent semantic dependencies are omitted. Methods that introduce semantics into then emerge, injecting knowledge from space where general language understanding compressed. However, existing semantic-enhanced focus on aligning two...
Recently, increasing attention has been paid to LLM-based recommender systems, but their deployment is still under exploration in the industry. Most deployments utilize LLMs as feature enhancers, generating augmentation knowledge offline stage. However, recommendation scenarios, involving numerous users and items, even generation with consumes considerable time resources. This inefficiency stems from autoregressive nature of LLMs, a promising direction for acceleration speculative decoding,...
Path planning plays an important role in the area of mobile robot, which includes two contents: environment modeling and path searching. The method Voronoi diagram for general figures by approximation structuring is utilized origin terminal are linked into to construct non-collision road map. Zonal Ant Colony Algorithm applied search shortest between terminal. Simulation results demonstrate that simple effective.
Based on the characteristic of mobile entity and obstacles, bounding box has been adopted to establish vector models, obstacles environment can be described more accurately. Basic obstacle avoidance theory spatial measurement algorithm are banded together improve efficiency for in a strait environment, over correctness feasibility validated by simulation experiments.
On‐orbit processors generally require reinforcement due to the harsh space radiation environment. Although traditional full triple modular redundancy (TMR) method can effectively reinforce circuits, it requires excessive physical footprints and energy consumptions. To address this problem, Letter proposes a general partial TMR that circuit. The is based on PageRank algorithm for sorting importance of triggers in circuit, mean time between failure dual logic cone model used as criterion...
The set pair evaluation method of surface warship formation operational capability has been proposed based on the analysis theory. basic ideas and steps are discussed. Compared with other methods, is clear in concept, simple to calculate convenient operation. Combined one example, this applied. result shows that a reliable effective method.