P. Kankuekul

About

P. Kankuekul is a researcher whose work sits at the intersection of computer vision, robotics, and machine learning, with a particular focus on enabling intelligent systems to learn continuously and adapt to new information. Their primary research areas include zero-shot learning, incremental learning, and transfer learning. Kankuekul’s most significant contribution is the development of an "Online incremental attribute-based zero-shot learning" method, a seminal paper with 105 citations that addresses a critical challenge in robotics and mobile communications: how to classify objects never seen during training. Unlike traditional offline batch methods, this work introduces a framework where attribute labeling is acquired through real-time user interaction, handling inconsistencies in the process. This approach allows robots to learn and adapt on the fly. Additionally, their work on "Fast online incremental transfer learning for unseen object classification" (10 citations) further advances the field by leveraging self-organizing incremental neural networks to transfer knowledge from known to unknown object classes. Kankuekul’s research is notable for its practical focus on real-world, dynamic environments, making their contributions highly relevant for developing autonomous systems that can learn and evolve without requiring complete retraining.

Research Focus

Key Achievements

Top Papers

1. 1
   Online incremental attribute-based zero-shot learning

   105 citations · 2012
2. 2
   Fast online incremental transfer learning for unseen object classification using self-organizing incremental neural networks

   10 citations · 2011

Key Collaborators

Contact & Links