Zheng Lian

Title: Innovations of Zheng Lian in Emotion Recognition Technology

Introduction

Zheng Lian is a prominent inventor based in Beijing, China. He has made significant contributions to the field of emotion recognition technology, holding a total of 5 patents. His work focuses on developing advanced methods for understanding and correcting emotional expressions in dialogues.

Latest Patents

Zheng Lian's latest patents include a "Dialogue emotion correction method based on graph neural network." This innovative method involves extracting acoustic, text, and image features from video files to create multi-modal features. It predicts the emotional tone of each sentence in a dialogue and refines these predictions by incorporating interaction information between speakers. This results in a more accurate emotion recognition outcome.

Another notable patent is the "Multimodal dimensional emotion recognition method." This method acquires frame-level audio, video, and text features from samples to be tested. It employs temporal contextual modeling using a temporal convolutional network to enhance the features before performing regression prediction to achieve a final dimensional emotion prediction result.

Career Highlights

Zheng Lian is affiliated with the Chinese Academy of Sciences, where he continues to push the boundaries of research in emotion recognition. His innovative approaches have garnered attention in both academic and industrial circles.

Collaborations

Zheng has collaborated with notable colleagues such as Jianhua Tao and Yu Wang, contributing to the advancement of their shared research interests.

Conclusion

Zheng Lian's work in emotion recognition technology exemplifies the potential of innovative methods to enhance human-computer interaction. His contributions are paving the way for more accurate emotional understanding in various applications.