Course details


Machine Learning in Cognitive Computational Neuroscience

L + S + Interdisciplinary Course (Master)
WS 2022 Prof. Dr. rer. nat. Tim Christian Kietzmann Hybrid
B.Sc modules:
CS-BWP-AI - Artificial Intelligence
CS-BWP-MCS - Methods of Cognitive Science
CS-BWP-NI - Neuroinformatics
KOGW-WPM-KI - Artificial Intelligence
KOGW-WPM-NI - Neuroinformatics
M.Sc modules:
CC-MP-IDC - Interdisciplinary Course
CC-MWP-AI - Artificial Intelligence
CC-MWP-NI - Neuroinformatics
CS-MP-IDC - Interdisciplinary Course
CS-MWP-AI - Artificial Intelligence
CS-MWP-NI - Neuroinformatics

CS-BW - Bachelor elective course
CS-MW - Master elective course
Doctorate program
Wed: 10-12
Mon: 10-12

Requirements: a basic understanding of neurobiology and neuropsychology (with a special focus on vision), as well as a solid understanding of deep learning and linear algebra. Abstract: Machine learning and neuroscience have a long intertwined history of trying to create and understand phenomena of intelligent, adaptive behaviour. The underlying connectionist approach has been highly influential for both fields. For machine learning, it led to the development of deep neural networks, which are built upon multiple biological inspirations (e.g. distributed coding, activation functions, stochasticity, dropout, attention, convolutions, etc). For computational neuroscience, neural networks act as a modelling framework for testing hypotheses of how distributed sets of simpler units can give rise to complex behaviour. In a true interdisciplinary fashion, these modelling efforts have benefited greatly from recent deep learning developments, as they allow for end-to-end trained systems performing complex tasks on real-world data. What emerged from this joint endeavour is the field of neuroconnectionism, which aims to integrate biological details into deep learning systems (architecture, learning objectives, and input statistics), while testing the resulting models against high-dimensional neural data and behaviour. This course introduces this interdisciplinary field, highlighting the many use cases of deep learning in modelling information processing in the brain. Students will learn (i) the rationale behind computational modelling in cognitive neuroscience, (ii) various deep learning techniques used to derive and adjust normative models of brain function (e.g. supervised, unsupervised, and reinforcement objectives, feedforward, and recurrent architectures, and meta learning), and (iii) multivariate analysis techniques to compare representations in biological and artificial neural networks (RSA, encoding models, behavioural testing, and in-silico experimentation).