易溶于水。 8 位比特币 gif。

I am currently a Ph.D. student with School of Eletronic and Information Engieering, South China University of Technology, China, where I obtained my bachelor degree. My research interest mainly lies in Natural Language Processing(NLP), Computer Vision(CV), Deep Learning(DL), Image Processing and Dialogue system.

I am advised by Prof. Xiangmin Xu and Associate Prof. Xiaofen Xing in SCUT and co-advised by Prof. Dacheng Tao in USYD.

Email: yirong.chen@outlook.com
LinkedIn: https://www.linkedin.com/in/chenyirong
Github: https://github.com/scutcyr
Publons: https://publons.com/researcher/3026890/yirong-chen
ORCID: http://orcid.org/0000-0002-0207-0067
ResearchGate: https://www.researchgate.net/profile/Yirong_Chen7
CSDN: https://blog.csdn.net/m0_37201243
Google Scholar: https://scholar.google.com/citations?user=aG5aDKEAAAAJ

我如何获得比特币地址

2019.09-Present South China University of Technology ICE，PhD

Main Experience: I served as the monitor of the 2019 level doctoral class in telecommunications of South China University of Technology and the president of the graduate association of Huazhou Student Union (2019). I was awarded the excellent graduate cadre of South China University of Technology in 2019-2020, the Xinwangda New Energy Scholarship in 2019-2020, and the outstanding graduate backbone of South China University of Technology in 2020-2021. Now I am the student leader of the Personalized and Emotional Conversation Group of the Engineering Research Center of the Ministry of Human Data Perception, So far, I have participated in three scientific research projects at or above the provincial and municipal levels related to the generation of personal emotional dialogue. During this period, I realized the standardized management of the laboratory server architecture and the unification of the research framework of emotion and personality interaction. As the first responsible person of students, I released the first Chinese personalized and emotional dialogue dataset CPED (Github won more than 80 stars in the first half of the year, and was also included in the luge.ai platform). I also proposed personality recognition in conversation (PRC) personalized and emotional conversation Generation (PEC) and other novel NLP tasks.

Skills: Pretrained Model, Open Domain Conversation, Python, Pytorch, Transformers, Bash, Ignite, Tensorflow, Linux operation and maintenance, server management, etc.

2015.09-2019.06 South China University of Technology E.E, B.E

Main Experience: I used to be the monitor of Electronic Science and Technology (Zhuoyue Class), the president of the general area of Guangdong Huazhou University Students Federation, and the director of the Hardware Group&ROS Group of the Robotics and Optical Communication Research Group of South China University of Technology.

Innovations: I have presided over or participated in 4 provincial/national scientific research projects (the National Undergraduate Innovation Training Program “Indoor Positioning of Service Robots Based on Visible Light Communication” won an excellent conclusion), published 4 SCI papers, applied for more than 10 invention patents, won 9 science and technology awards above the school level, won the first prize of Guangdong Undergraduate Electronic Design Competition, the first prize of 2017 American Undergraduate Mathematical Modeling Competition The first prize of the 2018 American Undergraduate Mathematical Modeling Contest, the first prize of the 2017 National Undergraduate Mathematical Modeling Contest (Guangdong) , the second prize of the 2016 National Undergraduate Mathematical Modeling Contest (Guangdong), and the silver medal of college students’ micro entrepreneurship action (20000 yuan).

Honors: I have won the first prize of the Xinmade Scholarship, the Huawei Scholarship (5000 yuan), the Huiding Science and Technology Special Scholarship (15000 yuan), the “Merit Student” of SCUT (twice), and the outstanding Communist Youth League members of SCUT. I led the team to win the “Advanced Class Collective” of SCUT, the “Top Ten Excellent Teams” of South China University of Technology, and the civilized dormitory of South China University of Technology (2017, 2018).

Skills: Mathematical modeling, STM32 programming, C language, robot ROS development, Java development, Android development, etc.

2012.09-2015.06 Huazhou No.1 Middle School

N/A

亚当支持比特币预测

Large Scale Chinese Personalized Emotional Dialogue DataSet

Period: 2019.09-2021.12

Considering the importance of personality, emotion and other information in robot human dialogue for dialogue generation, we independently built a dialogue personality emotion annotation tool, collected more than 12000 dialogues and more than 133000 sentences on 40 Chinese TV dramas, and provided three types of attribute annotation (surname, gender, age), five personality trait annotation, and two types of emotion annotation (three types of coarse grained emotion, 13 types of fine grained emotion), As well as the multi source knowledge related to emotion and personality, such as DA annotation of dialogue actions, we have won 44 stars in Github last month, and were also included in the domestic Qianyan platform. The constructed CPED dataset can be used to evaluate dialogue understanding tasks and dialogue generation tasks, such as speaker modeling, personality recognition in dialogue, emotion recognition in dialogue, DA recognition in dialogue, emotion dialogue generation, personality conversation generation, empathy dialogue generation, etc. It will play a positive role in promoting the development of cognitive intelligence. The paper has been submitted to TASLP (first author).

Study of Personality and Emotional Reasoning in Conversation

Period: 2021.09-2022.09

Considering the working scene of the service robot, understanding the speaker’s personality is of great significance for providing personalized services. In the dialogue, the speaker’s personality recognition task in 1-N dialogue is proposed, and the personality of a person is predicted based on his or her conversation with N other people. A model bert^c_{ssenet} is proposed. The F1 score of the three personality classifications in five dimensions reaches 74.08, and the accuracy rate reaches 67.25; Based on the speaker’s emotional consistency and mutation characteristics in conversation, sentence transfer relationship and emotion transfer relationship are modeled, and the conversation encoding emotion sequence decoding model CEED is implemented, which achieves good results in the conversation emotion recognition task and the conversation response emotion prediction task. The paper Speaker aware Hierarchical Transformer for Personality Recognition in Multiparty Dialogues has been submitted to ICASSP2023 (the second author).

Conversation Generation with Syntactic Structure Information

Period: 2021.01-2022.07

For the generation of personal emotional dialogue (including personalized conversation generation and emotional conversation generation), the implicit embedding {emo+da} – GPT of personality/emotion and the explicit embedding framework GPT – {per+emo+da} of personality/emotion are proposed, which further improves the personality expression and emotion expression ability of the pre training open domain dialogue generation model, and has a good improvement in PPL, Distinct-1/2, Bertscore and manual evaluation indicators, The ability of personified expression of the dialogue system has been significantly improved. This work has been submitted to TASLP together with the dataset CPED.

Conversation Generation with Syntactic Structure Information

Period: 2021.09-2022.07

According to the syntactic autonomy theory, information at different levels of sentences is completed by different functional modules of the brain during sentence processing, and syntactic processing is only affected by syntactic structure information. Therefore, sentence dependency is modeled at the self attention level, and sentence dependent attention DRA is proposed to effectively integrate dependency knowledge containing syntactic structure information into the Transformer block Further improve the humanoid expression performance of the multi round dialogue generation model on the EmpathicDialogues dataset. This achievement has been published in Workshop SUKI @ NAACL2022 (second author).

Prompt-based Multi-task Learning for Multimodal Conversation

Period: 2022.07-2022.11

A hint based multitask learning encoder decoder is proposed, in which different subtasks use different hints, so that the model tends to focus on the current subtask. He won the championship in sub task 1 (fuzzy candidate identification) of Track 1 of the 11th International Dialogue System Challenge (DSTC-11), and the runner up in sub task 2~4 (multi-mode coref resolution (MM Coref), multi-mode dialogue state tracking (MM-DST), and response generation). The paper has been submitted to DSTC-11 Workshop @ SIGDIAL 2023 (first author).