A5020312513- Advanced Malware Detection Techniques
- Adversarial Robustness in Machine Learning
- Distributed systems and fault tolerance
- Advanced Data Storage Technologies
- Anomaly Detection Techniques and Applications
- Video Analysis and Summarization
- Human Motion and Animation
- Parallel Computing and Optimization Techniques
- Caching and Content Delivery
- Multimedia Communication and Technology
- Advanced Data Compression Techniques
- Color perception and design
- Physical Unclonable Functions (PUFs) and Hardware Security
- Aesthetic Perception and Analysis
- Digital Media and Visual Art
- Image and Signal Denoising Methods
- Advanced Image Processing Techniques
Zhejiang University
2019-2024
Alibaba Group (China)
2023
Zhejiang University of Science and Technology
2020
Chung-Ang University
2012
Deep learning has been broadly leveraged by major cloud providers, such as Google, AWS and Baidu, to offer various computer vision related services including image classification, object identification, illegal detection, etc. While recent works extensively demonstrated that deep classification models are vulnerable adversarial examples, cloud-based detection models, which more complicated than classifiers, may also have similar security concern but not get enough attention yet. In this...
We present an automatic generation pipeline of interactive nonlinear video for online apparel shopping navigation. Our approach was inspired by Google's "Messy Middle" theory, which suggests that people mentally are faced with two tasks—exploration and evaluation—before purchasing online. Given a set product presentation videos, our navigation UI organizes them to optimize users' exploration automatically generates videos evaluation. To support methods, we proposed clustering similarity (...
Adversarial examples (AEs) against deep neural networks (DNNs) raise wide concerns about the robustness of DNNs. Existing detection mechanisms are often limited to a given attack algorithm. Therefore, it is highly desirable develop robust approach that remains effective for large group algorithms. In addition, most existing defences only perform well small images (e.g. MNIST and Canadian institute advanced research (CIFAR)) rather than ImageNet). this paper, authors propose defence method...
As a team from Alibaba Cloud, we have developed and open-sourced RocketMQ, cloud-native "messaging, eventing, streaming" real-time data processing platform that covers cloud-edge-device collaboration scenarios. During the development of also formulated RocketHA, log-based storage high availability design theory provides robust solution for distributed log software used in industrial applications. RocketHA comprises six fundamental components enable automatic cluster recovery failures such as...
Abstract The style of graphic design is an important factor that influences the memorability designs. Graphic designers routinely analyze latest designs, capture trends, and create designs match trends to appeal a larger audience. Nonetheless, lack quantitative analysis techniques can lead inefficient process introduce subjectivity bias. To expedite designers’ understanding make more objective, we propose GradeS, AI-driven support system facilitates multifaceted style. was designed developed...
In this paper, we present a deblocking algorithm that removes grid and staircase artefacts, which result from the compression of images. The KORea Multi-Purpose SATellite-2 images are efficiently compressed using an assigns quantisation coefficients, by rows, according to local region complexity, in such way more complicated regions compressed. However, approach is problematic it produces blocking artefacts those relatively less areas within same row as regions. Removing these with general...