Tutorial Session

Tutorial Title: Brain-inspired AI

Even though the contemporary artificial intelligence (AI) became a powerful tool in many areas and even in our everyday life, it still cannot outperform the human brain. The natural intelligence is a result of collaborative work of neural cell ensembles having much more elaborated functionality and connectivity than the artificial neural networks. The contemporary neuroscience revealed a lot about brain functioning principles and structural connectivity. The accumulated knowledge nowadays allows to develop brain-inspired models that would be able to solve variety of tasks in the way our brain does.

The main aim of this tutorial is to present to the potential listeners basic principles of how biological neural cells work, the existing brain-inspired neural cells models called spike timing neuron models and biologically inspired learning strategies like spike timing dependent plasticity (STDP) and reinforcement learning (RL). Examples of how our brain perceives and process visual information and takes decisions will be presented.

The practical part of the tutorial will include introduction to an open software - NEST Simulator - for spiking neural network (SNN) models. The listeners will learn how to create their own SNN structure using knowledge from neuroscience and how to simulate it using NEST Simulator library in Python.

Detailed outline of topics:

• Theme 1: Neural cells structure and spike timing neural cells models
• Theme 2: Learning in the brain: spike timing dependent plasticity
• Theme 3: Decision making in the brain: reinforcement learning and its neural correlates
• Theme 4: Examples of brain structures: visual system and basal ganglia
• Theme 5: NEST Simulator: basic neuron models; creating and simulation of SNN structures