Qiwei Wang

Qiwei Wang

Auto-generated profile, based on publicly available information

Qiwei Wang

School of Artificial Intelligence, National Key Laboratory for Novel Software Technology, Nanjing University, Nanjing, China

Research Focus & Metrics

Academic Career Length

6Years (2021-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
16 July 2024 in IEEE Transactions on Pattern Analysis and Machine Intelligence
Verified
Journal Article
Closed access
DZDa-Wei Zhou
Da-Wei Zhou
QWQi-Wei Wang
Qi-Wei Wang
ZQZhi-Hong Qi
Zhi-Hong Qi
HYHan-Jia Ye
Han-Jia Ye
DZDe-Chuan Zhan
De-Chuan Zhan
ZLZiwei Liu
Ziwei Liu
Deep models, e.g., CNNs and Vision Transformers, have achieved impressive achievements in many vision tasks in the closed world. However, novel classes emerge from time to time in our ever-changing world, requiring a learning system to acquire new knowledge continually. Class-Incremental Learning (CIL) enables the learner to incorporate the knowledge of new classes incrementally and build a universal classifier among all seen classes. Correspondingly, when directly training the model with new class instances, a fatal problem occurs — the model tends to catastrophically forget the characteristics of former ones, and its performance drastically degrades. There have been numerous efforts to tackle catastrophic forgetting in the machine learning community. In this paper, we survey comprehensively recent advances in class-incremental learning and summarize these methods from several aspects. We also provide a rigorous and unified evaluation of 17 methods in benchmark image classification tasks to find out the characteristics of different algorithms empirically. Furthermore, we notice that the current comparison protocol ignores the influence of memory budget in model storage, which may result in unfair comparison and biased results. Hence, we advocate fair comparison by aligning the memory budget in evaluation, as well as several memory-agnostic performance measures. The source code is available at https://github.com/zhoudw-zdw/CIL_Survey/ .

17 August 2023 in Journal of Education, Humanities and Social Sciences
Verified
Journal Article
Open access
QWQiwei Wang
Qiwei Wang
The quick advancement of artificial intelligence technologies represents a backdrop that cannot be disregarded in contemporary organizational management practice and research. Based on the field of human resource management, the study reviews and summarizes the impact of AI on the subject of human resource decision-making—employees—from the viewpoint of organizational management. This study takes two important research directions of AI and organizational management research at the individual level, and summarizes the research on employees' adoption of AI technology and the impact of AI on them individually in organizations. The adoption of AI technology by employees is influenced by technological, organizational and personal factors. AI technology shows the advantage of augmentation effect on employees on the one hand, and the phenomenon of automation, i.e. substitution effect on the other. Organizations should place a high value on the new changes that AI has made to the workplaces, attach importance to employees' attitudes, psychological and behavioral responses to new intelligent technologies. Utilize AI technologies to support decisions and promote the development of the new model of "AI+HRM".

1 December 2021
Verified
Conference Paper
Closed access
QWQiwei Wang
Qiwei Wang
Start from 2012, AI Technology was focused by the public again. As a sub-component, smart wearable devices are widely used with sensors for collecting the data about human body posture. Privacy and security problems are concerned by both users and developers in the using. Therefore, a reliable authentication technology with devices is necessary. This article is going to solve the problems of authentication using in the smart devices as smart bracelets. The article proposes an innovative method of smart bracelet authentication based on hand raising gesture and experiment the feasibility. Then, based on the research, a web-based program is designed and implemented using GIN framework to realize the bracelet authentication and user management. The sensors of the bracelets are used to record the three-axis accelerometer and the variation of the three-axis angular velocity of the arm movements. Features were extracted from the data set and classify them using a variety of algorithmic classifiers, such as multi-layer perceptron and random forest. Result indicators like TAR, FAR, and AUC are set to evaluate the effectiveness of the authentication method. In conclusion, authentication is feasible demonstrated by user identity web application but still need to be improved by diverse data in multiple environments.