1,322 results
Showing 1–20 · 0.40 seconds · RSS
Jisuanji kexue yu tansuo
Scopus
ISSN 1673-9418 · Journal of Computer Engineering and Applications Beijing Co., Ltd., Science Press · 1,322 articles in the index
-
Journal Article Full text
Class-Prototype Network and Dynamic Semantic Attention Module for Few-Shot Semantic Edge Detection
Accurate semantic edge information is crucial for machine vision applications such as object reconstruction, image generation, and medical imaging. However, existing deep learning-based semantic edge detection methods require a large amount of annotated data to train the network model, and the cost…
-
Journal Article Full text
Region and Channel Self-Replacement Image Data Augmentation Method
To address the limitations of existing data augmentation methods in enhancing model generalization, this paper proposes a novel approach named YNDP (you need double patch). The method selects a local patch from the image itself to replace other regions, constructing a new image with repeated…
-
Journal Article Full text
DynaSynergic-UNet: Dynamic Collaborative UNet for Medical Image Segmentation
The accuracy of medical image segmentation is crucial for disease diagnosis and treatment planning, yet existing methods still have limitations in multi-scale feature perception optimization, cross-level feature fusion alignment, and synergy between edge extraction and overall segmentation tasks…
-
Journal Article Full text
Multi-dimensional Spatial Perception and Global Feature Modulation for Robust RGBT Object Tracking
To address the susceptibility of RGBT object tracking to extreme lighting conditions and cluttered backgrounds in complex scenes, alongside existing methods’ inadequate modelling of shared structural information across modalities, this paper proposes an RGBT object tracking approach based on…
-
Journal Article Full text
Multimodal Jailbreak Attack Defense Method with Integrated Steering Vector Detection
With the widespread adoption of large vision-language models in multimodal tasks, their security faces severe challenges. Among them, jailbreak attacks bypass safety mechanisms through adversarial samples and carefully crafted prompts, inducing models to generate harmful content. Existing defense…
-
Journal Article Full text
Advances in Blockchain Privacy Protection Research
Blockchain technology has significant advantages in decentralization, but the contradiction between its public verifiability and the need for user privacy protection makes it a critical challenge to safeguard privacy during the data circulation process. This study analyzes blockchain-related…
-
Journal Article Full text
Open-Vocabulary Speech Emotion Recognition Based on Improved Low-Rank Adaptation
The emergence of audio large language models (AudioLLMs) has propelled speech emotion recognition (SER) from simple categorical discrimination toward profound emotional understanding, aiming to transcend current closed-set classification paradigms by generating diverse emotional descriptions beyond…
-
Journal Article Full text
Improved MADDPG Algorithm for Cooperative Pursuit via Joint Optimization of Experience Quality and Efficiency
To address training oscillations and slow policy convergence caused by poor experience quality and low utilization efficiency in multi-agent deep reinforcement learning (MARL) for cooperative pursuit tasks, an improved algorithm, STEP-MADDPG (sequence-temporal experience prioritized multi-agent…
-
Journal Article Full text
Review of Research on Knowledge Distillation Methods for Low-Resource Machine Translation
With the advancement of deep learning, neural machine translation models have achieved remarkable progress in translation tasks for high-resource language pairs and reached high levels on multiple automatic evaluation metrics. Low-resource machine translation, however, is often constrained by the…
-
Journal Article Full text
Long-Tail-Aware Semantic-Structural Decision Fusion for Entity Alignment
Entity alignment is a critical task of knowledge graph fusion, aiming to identify equivalent entities within different knowledge graphs. Although most existing related methods directly integrating structure and auxiliary information at the feature layer have achieved certain results, they have the…
-
Journal Article Full text
Survey on Reinforcement Learning-Based Cooling Control Methods for Green and Intelligent Buildings
Built-environment cooling systems face severe challenges in energy consumption and heat dissipation while ensuring thermal safety, so temperature regulation and constraint satisfaction are critical to operational efficiency and equipment longevity. Traditional physics- and rule-based controllers…
-
Journal Article Full text
Privacy-Preserving Localization Based on Secure Multi-party Computation
Location-based services serve as the foundation for applications in various fields such as smart cities, traffic navigation, and military reconnaissance. However, existing positioning methods based on satellites, RFID, and WLAN often suffer from issues like location privacy leakage for both users…
-
Journal Article Full text
YOLO-Agent: Multi-scale Lightweight Detection and Agent-Driven Decision-Making for Dairy Cattle Behavior
As the intelligent transformation of animal husbandry accelerates, large-scale dairy farms have an urgent need for automated behavior monitoring and intelligent decision-making. Existing deep-learning-based detection methods still suffer from insufficient robustness under complex farm conditions…
-
Journal Article Full text
Domain Multimodal Neural Machine Translation Method with Visual Adaptive Segmentation Contrastive Enhancement
Domain-specific multimodal neural machine translation (DMNMT) incorporates images as an additional modality input to achieve accurate translation from the source language to the target language within specific domains, thereby enhancing translation accuracy and robustness. Since visual information…
-
Journal Article Full text
From Language to Action: Survey of LLM-Driven Robotic Manipulation Paradigms
Traditional robot control in unstructured environments is often hindered by programming complexity, limited generalization, and heavy reliance on domain experts. This paper provides a systematic review of robot control empowered by large language models (LLMs), summarizing recent progress and major…
-
Journal Article Full text
Efficient Jailbreaking Attack via Experience Reuse and Coordinated Optimization
With the widespread adoption of large language models, their security vulnerabilities have become increasingly evident. Adversarial jailbreaking attacks, such as those based on the greedy coordinate gradient (GCG) method, can bypass safety alignment mechanisms and induce models to generate harmful…
-
Journal Article Full text
Adaptive Physics-Informed Neural Network Based on Dual-Indicator Sampling Strategy
An adaptive hybrid sampling method driven by dual indicators (residual and residual gradient), named DI-HAS (dual-indicator driven hybrid adaptive sampling), is proposed to address the optimization of collocation point sampling strategies in the training of physics-informed neural networks (PINNs)…
-
Journal Article Full text
Survey of Information Bottleneck Frameworks in Multimodal Representation Learning
Multimodal learning confronts multiple fundamental challenges, including modality heterogeneity, information redundancy, missing modalities, and computational efficiency. The information bottleneck (IB) theory, through its mathematical framework that optimally balances information compression and…
-
Journal Article Full text
Unified Evaluation Benchmark for Temporal Knowledge Graph Representation Learning
Temporal knowledge graph representation learning (TKGR-LE) aims to embed dynamic facts and entity relationships into a low-dimensional vector space to capture the semantic evolution along the time dimension. This technique is widely applied in various scenarios, such as financial risk control…
-
Journal Article Full text
DKVMN-PEPGen: Personalized Exam Paper Generation Model Based on Deep Knowledge Tracking
To address the limitations of existing automatic test generation research, which largely produces standardized papers that fail to consider individual differences in students?? knowledge states and often neglects question diversity, leading to similar or repetitive questions, this paper aims to…
Showing at most 50 pages. Narrow your search with filters for more specific results.