LS 04-Networks, learning, sim

Biological learning in neural networks

LS - School of Life Sciences

Computational Neuroscience

Computational Neuroscience

BiologyBiotechnologyComputer Science/ InformaticsElectrical EngineeringMathematicsPhysics

Participation also possible online only:

Planned project location:

School of Life Sciences, Freising

Prof. PhD

Julijana

Gjorgjieva

gjorgjieva@tum.de

+49 (8161) 71 - 2709

Dr.

Dylan

Festa

dylan.festa@tum.de

Elizabeth

Herbert

elizabeth.herbert@tum.de

+49 (8161) 71 - 2136

Neural networks in the brain are very different from artificial neural networks. They exhibit complex dynamics characterized by non-random connectivity that underlies our ability to perform different actions and behaviors, to think and remember things. Our group investigates how these dynamics are established and how networks continually change during learning. We specifically focus on diverse mechanisms of synaptic plasticity – the process by which the connections between neurons in a network are updated during ongoing neural activity. We identify these mechanisms from experimental data and study their role in computational models of learning. Our specific interest is in learning during development, when brains are very immature and networks self-organize into their adult counterparts. Some current active routes for research include: studying computational properties of emerging network structures resulting from different plasticity rules, analyzing the effect of spontaneous activity in shaping neural circuit structure in early developmental stages right after an animal is born, and modeling top-down contextual modulation of neuronal circuits with multiple interneuron types. We mainly draw inspiration from experimental data collected by our collaborators, and build theoretical models that pertain mainly to (mammalian) sensory cortices.

Your research project will involve working within an interdisciplinary group of scientists including biologists, engineers, mathematicians, and physicists, actively collaborating with several members of our research group working on the frontiers of Computational and Theoretical Neuroscience. The exact details of the project depend on both the student background and interest but will aim to investigate the interplay of neuronal and synaptic properties on the computations and behaviors performed by brain networks on developmental and/or evolutionary timescales. For example, what do different neuronal types contribute to network computations? How do networks stabilize their connectivity architectures during learning? How can neuronal activity be correlated to behavior?

Depending on interest, tasks will involve statistical/data analysis, computational modeling, analytical calculations, and finally preparing an oral presentation and a written report. The project may result in a peer reviewed publication. During your stay here you will have the opportunity to participate and actively engage in discussing science in our weekly lab meetings and bi-weekly journal clubs.

Mon-Fri, ca. 30-40 hrs/week

2 years of bachelor studies completed

Applicants will ideally have a strong background in exact sciences (e.g. biology, mathematics, physics, engineering and related disciplines) with an interest in Computational/Theoretical Neuroscience. Applicants should know the basics of linear algebra and differential equations.

Experience with data analysis and programming in Matlab, Python, Julia, C++ or another programming language is strongly recommended.