Max Robotics — Global Service Robot Directory + US Certification

About

Shogo Nagasaka is a leading researcher in developmental robotics and computational cognitive science, whose work focuses on how machines can autonomously acquire language and concepts in a manner inspired by human infants. His core research areas include multimodal concept formation, unsupervised word discovery, and the integration of perception and language. Nagasaka’s most impactful contribution is the development of online algorithms for multimodal categorization, most notably through his work on multimodal latent Dirichlet allocation (MLDA) and hierarchical Pitman-Yor language models. His 2012 paper on online learning of concepts and words using these models has garnered 46 citations, establishing a foundational method for robots to learn from both sensory data and partial human input. He further advanced this line of inquiry in 2014 with a study on mutual learning of object concepts and language models (44 citations), demonstrating how robots can simultaneously refine their understanding of objects and the words that describe them. Nagasaka has also tackled the challenging problem of direct word discovery from raw speech signals, proposing a double articulation analyzer with deep sparse autoencoders (29 citations). His work is notable for its constructive approach to understanding human developmental capabilities, offering a computational bridge between infant learning and artificial intelligence.