Cognitive Neurorobotics
This course aims to provide an overview of the synthetic approach to understand embodied cognition by using deep dynamic neural network models and robotics platforms.
Explore the principles of embodied cognition by a synthetic neurorobotics modeling approach in combination with hands-on neurorobotics experiments and related term projects. Combine related interdisciplinary findings in artificial intelligence and robotics, phenomenology, cognitive neuroscience, psychology, and deep and dynamic neural network models. Perform neurorobotics simulations and control experiments with extensive coding in C++ or Python. Critically analyze and report on recent papers in neurorobotics and artificial intelligence.
1. Introduction: cognitive neurorobotics study
2. Cognitism: compositionality and symbol grounding problem
3. Phenomenology: consciousness, free will and embodied minds
4. Cognitive neuroscience I: hierarchy in brains for perception and action
5. Cognitive neuroscience II: Integrating perception and action via top-down and bottom-up interaction
6. Affordance and developmental psychology
7. Nonlinear dynamical systems I: Discrete time system
8. Nonlinear dynamical systems II: Continuous time system
9. Neural network model I: 3-layered perceptron, recurrent neural network
10. Neural network model II: deep learning, variational Bayes
11. Neurorobotics I: affordance & motor schema
12. Neurorobotics II: higher-order cognition, meta-cognition, and consciousness
13. Neurorobotics III: hands-on experiments in lab
14. Paper reading for neurorobotics and embodied cognition I
15. Paper reading for neurorobotics and embodied cognition II
Mid-term exam: 40%, final term project report: 60%.
B46 Introduction to Machine Learning and programming experience in Python, C or C++ are required. Basic calculus of vectors and matrices and differential equations are assumed.
Exploring robotic minds: actions, symbols, and consciousness as self-organizing dynamic phenomena. Jun Tani (2016) Oxford University Press.