C. James Li

About

C. James Li is a researcher whose work sits at the intersection of robotics, control systems, and neural network-based learning — a field that was particularly pioneering in the 1990s when intelligent machine control was still in its formative stages. Li's most influential contribution, cited 37 times, introduced automatic methods for training recurrent neural networks (RNNs) to model the structure and parameters of mechanical systems, significantly advancing how machines could learn dynamic behavior without exhaustive manual specification. His 1997 work on robot learning control demonstrated the practical power of RNN-based inverse dynamics modeling, enabling robots to operate effectively even when their underlying dynamics are initially unknown — a landmark step toward adaptive, self-learning robotic systems. Li also made foundational contributions to neural network design itself, developing the Augmentation by Training with Residuals (ATR) algorithm, which removed the burden of manual configuration by automating both structural and weight learning in feedforward networks. Taken together, his body of work reflects a sustained commitment to making intelligent control systems more autonomous, accessible, and applicable to real-world engineering challenges — contributions that continue to resonate in modern robotics and machine learning research.

Research Focus

Key Achievements

Top Papers

1. 1
   Automatic structure and parameter training methods for modeling of mechanical systems by recurrent neural networks

   37 citations · 1999
2. 2
   Robot learning control based on recurrent neural network inverse model

   22 citations · 1997
3. 3
   A New Feedforward Neural Network Structural Learning Algorithm—Augmentation by Training With Residuals

   11 citations · 1995

Key Collaborators

Contact & Links