Master Thesis AVIAN: AVian Inspired Artificial Network for Vocal Production

alternate text Figure 1. Model circuitry and brain recordings. Left: Simplified songbird motor control system with key components. Right: Neural activity during singing in the variability generating premotor pathway in songbirds. Recorded data are to be used as in and output to train and evaluate the model.

Research context

How can neural spiking activity translate into behavioral variants of a complex behavior? We study this question in songbirds that have evolved a discrete and modular brain circuitry for song learning and production. Central to this circuitry are two converging premotor pathways: one driving stereotyped behavior and providing temporal information, and another introducing behavioral variability [1].

 

Master thesis

The goal of this master thesis is to develop a bio-inspired neural network modeled after the songbird's motor control system (Figure 1). This network emulates the two pathways including diverse receptor dynamics [2] and utilizes large-scale neural recordings from the variability-generating premotor pathway to map spiking activity to behavioral irregularities. Ultimately, the project aims to uncover biological mechanisms and constraints that are challenging to explore through direct experimentation.

 

Tasks

As part of this thesis, you will:

  • - Review existing literature on related neuroscience and machine learning topics.

  • - Develop and implement a bio-inspired neural network.

  • - Evaluate the network's performance using various bio-inspired activation functions [3]

  • - Document your code, results, and methods comprehensively.

  • - Write a final report summarizing your findings.

We provide weekly meetings to discuss outcomes, generate ideas, and define next steps. The workload is designed for six months of full-time commitment.

 

Your benefits 

You will learn about neuroscience, biophysics, signal processing, and modern ML, gaining expertise in scientific use of neural network and data analysis methods and work in an interdisciplinary group. 

 

Your profile 

We are looking for a highly motivated student with interests in neuroscience, machine learning and modelling, and programming. Ideal candidates should have a strong academic background, enthusiasm for interdisciplinary research, and proficiency in relevant technical skills.

To apply, please send your CV and transcript of records to one of the contacts below.

 

Contacts: 

Dr. Corinna Lorenz: corinna@ini.ethz.ch

Prof. Dr. Richard Hahnloser: rich@ini.ethz.ch

 

Literature

[1] Fiete & Seung, 2007, Elsevier

[2] Stark and Perkel, 1999, JNeurosc

[3] Kim et al., 2024, NeurIPS

back
back