SAN JOSE, Calif., Feb. 27, 2026 — The University of San Jose (USJ) hosted a distinguished academic lecture on artificial intelligence research as part of its USJ Academic Lecture Series, welcoming leading AI expert Dr. Shaojun Wang to present his latest work on large language model (LLM) reasoning.
The lecture, titled “Search PPO with Shared Actor-Critic for LLM Reasoning,” explored new approaches to improving reasoning capabilities in large language models through reinforcement learning techniques. The event brought together faculty members, researchers, technology professionals, and students interested in the rapidly evolving field of AI.

During the presentation, Dr. Wang introduced Search PPO, a principled extension of Proximal Policy Optimization (PPO) that integrates likelihood-based beam search into reinforcement learning training at the token level. The approach aims to improve exploration during training and enable large language models to produce higher-quality responses for complex reasoning tasks.
Dr. Wang also discussed a shared actor–critic architecture built upon a large language model backbone. The framework incorporates a Transformer-based value head, enabling scalable reinforcement learning fine-tuning for long-horizon reasoning problems. According to Dr. Wang, this architecture offers a promising pathway toward enhancing reasoning performance in next-generation AI systems.

“Advances in reinforcement learning are opening new possibilities for improving the reasoning abilities of large language models,” Dr. Wang said during the lecture. “By combining search methods with actor–critic training, we can better guide model optimization and achieve more reliable reasoning outcomes.”
The event concluded with an interactive discussion session, where attendees engaged with the speaker on topics including LLM optimization, reinforcement learning methodologies, and emerging trends in artificial intelligence research.
“Hosting distinguished scholars and industry leaders is an important part of USJ’s mission to promote academic exchange and innovation,” said Dr. Claude Wang, founder of the University of San Jose. “We are committed to building a vibrant platform where researchers and students can explore the latest advances in artificial intelligence and emerging technologies.”

Dr. Shaojun Wang currently serves as Chief Natural Language Processing Scientist at Ping An Technology and Head of the Speech and Language Laboratory. He holds bachelor’s and master’s degrees in Electrical Engineering from Tsinghua University and a Ph.D. in Electrical and Computer Engineering from the University of Illinois Urbana–Champaign (UIUC).
The lecture is part of the USJ AI Seminar Series, an ongoing initiative designed to bring leading researchers and innovators to campus and foster collaboration in cutting-edge technology fields.
