Jia-yin QI

Dr. Jia-yin QI


Research Focus & Metrics

Academic Career Length

7Years (2020-Present)

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

Representative Publications

View All Publications
21 July 2022 in Entropy
Verified
Journal Article
Open access
PDF file
Feng LiuFeng Liu
Feng Liu
SSSi-Yuan Shen
Si-Yuan Shen
ZFZi-Wang Fu
Zi-Wang Fu
HWHan-Yang Wang
Han-Yang Wang
AiMin ZhouAiMin Zhou
AiMin Zhou
Jia-yin QIJia-yin QI
Jia-yin QI
Semantic-rich speech emotion recognition has a high degree of popularity in a range of areas. Speech emotion recognition aims to recognize human emotional states from utterances containing both acoustic and linguistic information. Since both textual and audio patterns play essential roles in speech emotion recognition (SER) tasks, various works have proposed novel modality fusing methods to exploit text and audio signals effectively. However, most of the high performance of existing models is dependent on a great number of learnable parameters, and they can only work well on data with fixed length. Therefore, minimizing computational overhead and improving generalization to unseen data with various lengths while maintaining a certain level of recognition accuracy is an urgent application problem. In this paper, we propose LGCCT, a light gated and crossed complementation transformer for multimodal speech emotion recognition. First, our model is capable of fusing modality information efficiently. Specifically, the acoustic features are extracted by CNN-BiLSTM while the textual features are extracted by BiLSTM. The modality-fused representation is then generated by the cross-attention module. We apply the gate-control mechanism to achieve the balanced integration of the original modality representation and the modality-fused representation. Second, the degree of attention focus can be considered, as the uncertainty and the entropy of the same token should converge to the same value independent of the length. To improve the generalization of the model to various testing-sequence lengths, we adopt the length-scaled dot product to calculate the attention score, which can be interpreted from a theoretical view of entropy. The operation of the length-scaled dot product is cheap but effective. Experiments are conducted on the benchmark dataset CMU-MOSEI. Compared to the baseline models, our model achieves an 81.0% F1 score with only 0.432 M parameters, showing an improvement in the balance between performance and the number of parameters. Moreover, the ablation study signifies the effectiveness of our model and its scalability to various input-sequence lengths, wherein the relative improvement is almost 20% of the baseline without a length-scaled dot product.

25 May 2022 in Applied Sciences
Verified
Journal Article
Open access
PDF file
Feng LiuFeng Liu
Feng Liu
ZFZhefu Feng
Zhefu Feng
Jia-yin QIJia-yin QI
Jia-yin QI
Assets such as warehouse receipts are important for enterprises, which can be used to pledge in supply chain finance (SCF). However, traditional pledges are performed manually, which inevitably encounters inefficiency and security problems such as multiple pledges. To improve asset security, we propose a blockchain-based digital asset platform (BDAP) with multi-party certification. BDAP not only has a security protocol based on the threshold ECDSA algorithm to make related participants confirm the authenticity of assets but also embeds a Verifiable Byzantine Fault Tolerant (VBFT) mechanism, randomly selecting the consensus nodes and improving the safety of the nodes. Moreover, data stored on the blockchain makes traceability possible. Through a set of experiments, we have verified the functionality and performance of BDAP. When the pressure test reaches 100 concurrent user volume, BDAP’s average response time is 1.441 s, showing a high ability to process transactions. However, now just a few open-minded banks are willing to access BDAP, it might take a long time to change the traditional perception of the participants in supply chain finance.

ZZZhenping Zhang
Zhenping Zhang
HYHaiyan Yan
Haiyan Yan
Jia-yin QIJia-yin QI
Jia-yin QI
Entrepreneurship is becoming increasingly essential in this current era of the knowledge economy. It contributes to the innovation of products and services as well as improved processes. In the long-run, it can also improve the sustainability of the economy by depicting better efficiency and social goals. To stimulate entrepreneurship, it is essential to investigate the thinking behind entrepreneurship or what entrepreneurs think about entrepreneurship. Such investigations should encompass the mental images of entrepreneurs. In this regard, content analysis, based on the popular Zhisland essays, may be applied to elicit opinions from Chinese entrepreneurs about activities, critical factors and intended outcomes within the ambit of entrepreneurship. In this study, 634 concepts are first coded and categorized into 20 second-level themes and six first-level themes. The six first-level themes are competing strategy, human resource management, management and leadership, marketing and sale, research and development, and risk management. Furthermore, among the 20 second-level themes, leadership, self-improvement, and the risk of business cycle attract the highest attention, each of which accounts for around 10% of the coded concepts. Finally, a causal loop diagram is depicted to synthesize the coding results. This study also underscores three essential activities of entrepreneurship, which entail building and maintaining competing advantage, improving user experience, and risk management. Entrepreneurs need to balance investments in those activities according to the change in environment and customers’ needs.

7 December 2022
Verified
Book Chapter
Closed access
PDF file
YLYangyang Lv
Yangyang Lv
SHShuaibo Hu
Shuaibo Hu
Feng LiuFeng Liu
Feng Liu
Jia-yin QIJia-yin QI
Jia-yin QI