Renato Ramos da Silva

About

Renato Ramos da Silva is a researcher at the intersection of developmental robotics and machine learning, with a primary focus on endowing social robots with human-like learning capabilities. His work centers on the challenging problem of **joint attention**—a fundamental non-verbal communication skill that humans naturally acquire in early childhood. Da Silva’s major contribution lies in applying **relational reinforcement learning** to model this shared attention, enabling robots to learn from human tutelage rather than through rigid, pre-programmed instructions. His most cited paper, "Modelling Shared Attention Through Relational Reinforcement Learning" (2011, 14 citations), explores how robots can learn to coordinate attention with a human partner by using relational representations, overcoming limitations of traditional Markov Decision Process models. In earlier work, "Concept Learning By Human Tutelage For Social Robots" (2008, 6 citations), he demonstrated how robots could acquire concepts through interactive teaching. Da Silva’s research is particularly notable for addressing the **state classification problem** in reinforcement learning, proposing hybrid architectures that combine relational reinforcement learning with recurrent neural networks to solve joint attention tasks. While his citation counts reflect a specialized niche, his work represents an important step toward more socially intelligent robots capable of natural, intuitive human-robot interaction.

Research Focus

Key Achievements

Top Papers

1. 1
   Modelling Shared Attention Through Relational Reinforcement Learning

   14 citations · 2011
2. 2
   Concept Learning By Human Tutelage For Social Robots

   6 citations · 2008
3. 3
   Relational reinforcement learning and recurrent neural network with state classification to solve joint attention

   4 citations · 2011

Key Collaborators

Contact & Links