Dienstag, 15.12.2020, 10.00 Uhr
Learning and Decision Making in Closed Loop Simulations of Plastic Spiking Neural Networks
- Zoom: https://rwth.zoom.us/j/99233095930?pwd=dHhTV253V1ZYUzRtSkk1L3A1REZVUT09
- Referent: Philipp Weidel, Dipl. Inform.
To understand how animals and humans learn, form memories and make decisions is a highly relevant goal both for neuroscience and for fields that take some inspiration from neuroscience, such as machine learning and artificial intelligence. Many models of learning and decision making were developed in the fields of machine learning, artificial intelligence, and computational neuroscience. Although these models aim to describe similar mechanisms, they do not all pursue the same goal. These models can be differentiated between models aiming to reach optimal performance on a specific task (or set of tasks) and models trying to explain how animals and humans learn. Some models of the first class use biologically inspired methods (such as deep learning) but are usually not biologically realistic and are therefore not well suited to explain the function of the brain. Models in the second class focus on being biologically plausible to explain how the brain works, but often demonstrate their capability on too simplistic tasks and yield low performance on well-known tasks from machine learning. This work aims to close the gap between these two types of models.
In the first part of this talk, tools are described that allow the combination of biologically plausible neural network models together with powerful toolkits known from machine learning and robotics. To this end, MUSIC, the middleware for spiking neural network simulators such as NEST and NEURON is interfaced with ROS, a middleware for robotic hardware and simulators such as Gazebo. This toolchain is extended with interfaces to reinforcement learning toolkits such as the OpenAI Gym.
The second part addresses the question of how the brain can represent its environment in the neural substrate of the cortex and how a realistic model of reinforcement learning can make use of these representations. To this end, a spiking neural network model of unsupervised learning is presented which is able to learn its input projections such that it can detect and represent repeating patterns. By using an actor-critic reinforcement learning architecture driven by a realistic dopamine modulated plasticity rule the model can make use of the representations and learn a range tasks.
Es laden ein: die Dozentinnen und Dozenten der Informatik