
Wei Peng
evaluation
education
stance detection
dialogue summarization
natural language generation
task-oriented dialog
large language models
meeting summarization
controllable text generation
hallucinations
text recognition
self-distillation
standards
hybrid knowledge management
analysis
11
presentations
3
number of views
1
citations
SHORT BIO
I am passionate about General-purpose AI, and I solve AI by drilling into two dimensions: 1) design and develop intelligent agents based on interdisciplinary theories of human cognition and deep learning and 2) through the application of AI, drive AI research to serve the practice and, in turn, pose new challenging research questions to advance AI. I am a leading expert in AI and cognitive systems, with over 15 years of extensive experience leading AI research and development in AI labs of Fortune 100 companies and research institutions. Based on the top-notch capabilities cultivated in serving the academic and industrial world, I have talents in driving AI research and landing AI technologies in commercial products. As a Director and Principal AI Scientist at the AI Application Research Center, Huawei Technologies, I lead a team of 50+ research scientists and engineers in developing general-purpose AI services (OCR, Speech Machine Translation, Knowledge-enhanced QA, etc.), resulting in hundreds of million calls in 2022. During the past four years, I have led the team to win championships in domain-specific machine translation shared tasks in WMT. I have in-depth insights into bridging the gap between scientific research and commercial product development with a strong track record of excellence in landing AI techniques to multiple industrial domains, including targeted sales, medical practice, smart grids, engineering design, and the financial sector. Prior to joining Huawei, I was a senior lecturer of business analytics at La Trobe University (2016-19), a senior data scientist at Telstra, and a technical lead data scientist at Lenovo. I was a senior data mining analyst (EL1) at AUSTRAC from 2010 to 2013, applying machine learning to combat money laundering and other organized financial crimes. I was a research fellow with RMIT University (2008-10) and an honorary adjunct principal research fellow (2011-14) focusing on applied artificial intelligence in smart grid and knowledge-based engineering after my postdoctoral research works with CSIRO ICT Centre. I regularly serve on academic community services as being various AI conference program committees, including ACL, EMNLP, ACL, AAAI, etc. I publish 50+ peer-reviewed papers in peer-reviewed AI conferences and journals and am the inventor of multiple patents.
Presentations

Enhancing Semantic Consistency of Large Language Models through Model Editing: An Interpretability-Oriented Approach
JINGYUAN YANG and 5 other authors

Assessing Factual Reliability of Large Language Model Knowledge
Weixuan Wang and 3 other authors

Self-Distillation Regularized Connectionist Temporal Classification Loss for Text Recognition: A Simple Yet Effective Approach
Ziyin Zhang and 6 other authors

Learning from Failure: Improving Meeting Summarization without Good Samples
Ke Wang and 2 other authors

ORCHID: A Chinese Debate Corpus for Target-Independent Stance Detection and Argumentative Dialogue Summarization
Xiutian Zhao and 2 other authors

M3Seg: A Maximum-Minimum Mutual Information Paradigm for Unsupervised Topic Segmentation in ASR Transcripts | VIDEO
Ke Wang and 3 other authors

PROSE: A Pronoun Omission Solution for Chinese-English Spoken Language Translation
Ke Wang and 3 other authors

Huawei BabelTar NMT at WMT22 Biomedical Translation Task: How We Further Improve Domain-specific NMT
Wei Peng and 1 other author

Neural Machine Translation with Heterogeneous Topic Knowledge Embeddings
Wei Peng

Huawei AARC's Submissions to the WMT21 Biomedical Translation Task: Domain Adaption from a Practical Perspective
Wei Peng

HyKnow: End-to-End Task-Oriented Dialog Modeling with Hybrid Knowledge Management
Silin Gao and 4 other authors