A5018049738- Metaheuristic Optimization Algorithms Research
- Human Pose and Action Recognition
- Image Enhancement Techniques
- Advanced Image and Video Retrieval Techniques
- Network Security and Intrusion Detection
- Hand Gesture Recognition Systems
- 3D Shape Modeling and Analysis
- Face and Expression Recognition
- Adversarial Robustness in Machine Learning
- Face recognition and analysis
- Evolutionary Algorithms and Applications
- Computer Graphics and Visualization Techniques
- Remote Sensing and LiDAR Applications
- Artificial Immune Systems Applications
- Gait Recognition and Analysis
- Anomaly Detection Techniques and Applications
Beijing University of Posts and Telecommunications
2020-2025
Multi-level thresholding is one of the essential approaches for image segmentation. Determining optimal thresholds multi-level needs exhaustive searching which time-consuming. To improve efficiency, a novel population based bee foraging algorithm (BFA) presented in this paper. The proposed provides different flying trajectories types bees and takes both single-dimensional multi-dimensional search aiming to maintain proper balance between exploitation exploration. swarm divided into number...
Scene graphs of point clouds help to understand object-level relationships in the 3D space. Most graph generation methods work on 2D structured data, which cannot be used for unstructured cloud data. Existing point-cloud-based generate scene with an additional structure that needs labor-intensive manual annotation. To address these problems, we explore a method convert into data and without given structures. Specifically, cluster points similar augmented features groups establish their...
Human motion prediction has achieved a brilliant performance with the help of convolution-based neural networks. However, currently, there is no work evaluating potential risk in human when facing adversarial attacks. The attack will encounter problems against naturalness and data scale. To solve above, we propose new method that generates worst-case perturbation by maximizing predictor's error physical constraints. Specifically, introduce novel adaptable scheme facilitates to suit scale...
Intrusion detection system (IDS) is a security solution to detect the anomalies of Internet Thing (IoT) network and raise alerts. Feature selection one essential preprocess steps build an efficient intelligent intrusion (IDS), which helpful reduce number features (NF) that are used classify attacks. The meta-heuristic algorithm-based approaches has been widely solve feature problems in past few years. However, as data dimension increases, dramatically expanded search space challenges...