Abstract: In recent years, large language models (LLMs) have been crushing it in complex reasoning challenges, with OpenAI's latest o-series and DeepSeek's R1 architecture leading the charge in deep reasoning technology. This talk is all about Scientific Discovery in the LLMs Era. We're going to take a deep dive into the core techniques behind knowledge reasoning in these massive models. We'll explore how to effectively empower LLMs with knowledge reasoning capabilities—through methods such as Chain-of-Thought, reasoning with reinforced fine-tuning, process supervision, self-reflection, critique-based supervision, and cross-modal reasoning. The goal? To equip LLMs with next-level reasoning powers and lay the groundwork for an awesome future in Scientific AI!
Bio: Professor Xuanjing Huang is a Professor at the School of Computer Science and Artificial Intelligence, Fudan University in Shanghai, China. As the head of the Fudan NLP Lab, her research is centered on natural language processing and large language models, a field in which she has published over 200 papers in major NLP, AI, and IR conferences and journals, accumulating over 30,000 citations according to Google Scholar. Her groundbreaking work has been widely recognized with multiple accolades, including best paper awards at CCL 2019, CCL 2023, and Instruction Workshop @ NeurIPS 2023, an area chair favorites award at COLING 2018, and outstanding paper awards at ACL 2017, NLPCC 2019, and COLING 2022. Beyond her research, she is deeply committed to community service, currently serving as an executive committee member at ACL, the AACL chair, and a steering committee member of SIGIR-AP, having previously demonstrated leadership as a Program Co-Chair for SIGIR-AP 2023, EMNLP 2021, CCL 2019, and NLPCC 2017, and as a General Co-Chair for NLPCC 2020.