Frederic Theunissen, Professor

Closed (1) Machine Learning Recognition of Animal Calls

Check back for status

You might have heard of ASR: Automatic Speech Recognition that is performed by Alexa or Siri. Our laboratory studies vocal communication in animals and would like to develop an automatic classifier for bird calls using advanced machine learning techniques.

The URAP will be writing Python code to implement supervised classification algorithms that will decode the call type and caller id for vocalizations produced by zebra finches and canaries. In addition to further gaining expertise in machine learning techniques, the URAP will also signal processing methods used for sound analyses and become familiar with current topics in animal communication.

Day-to-day supervisor for this project: Kevin Yu, Graduate Student

Qualifications: Undergraduate research apprentices need to have significant programming experience using Python, including the use of scikit-learn and Jupiter notebooks. Experience with sound analyses techniques is a plus. Fun position for CS majors who want to be involved in an applied project. Also great for biologists interested in animal communication and have some machine learning experience.

Weekly Hours: 6-8 hrs

Off-Campus Research Site: This is a remote project that can be performed at home.



Closed (3) Histological Techniques for Assisting Multiple Neuroscience Projects

Check back for status

The apprentice(s) will prepare and analyze neural tissue to assist graduate students and staff scientists with ongoing projects. This will include slicing, mounting (on microscope slides), and imaging of brain tissue to verify where electrodes were placed during electrophysiological recordings, as well as other projects. Apprentice(s) will learn some neuroanatomy and a lot of histological techniques.

The apprentice will:
- prepare brains, slice brains, mount them on slides, and stain and image them
- antibody and NISSL staining and quantification of brain regions and lesion sizes
- verification of electrode placement following electrophysiological recordings

The apprentice will spend ~6-8 hours weekly in the lab working on this project.

The apprentice will be given a reading list and will be invited to attend weekly lab meetings. At the end of the term, the apprentice will complete a final report as per the URAP contract and an 8-10 page term paper if obtaining URAP credit.


Day-to-day supervisor for this project: Bill Wood, Staff Researcher

Qualifications: Introductory courses completed in Psychology, Cognitive Science, or Biology (MCB or IB), and preferably enrolled in upper-division classes in any of these areas. Preference for students with a strong interest in neuroscience/neurobiology and enrolled in either the Psychology, Cognitive Science, MCB or IB majors. Previous lab experience is preferred but not required. Programming experience preferred but not required. Attention to detail and ability to work with computers and software are a must. Daily access to e-mail, and the ability to work independently are desired.

Weekly Hours: 6-8 hrs

Related website: http://theunissen.berkeley.edu/

Closed (4) Perception of Communication Calls in the Zebra Finch

Check back for status

Project Description:
The zebra finch has a very rich vocal repertoire of communication calls that are used in distinct behavioral contexts: peer recognition and localization, mate interactions, offspring care, etc. We previously gathered a huge bank of these calls, classifying them along semantic categories, i.e. groups of calls with the same meaning, as inferred from the observation of birds' behavioral interactions when they vocalize. We have now embarked on 'cracking the code' of this bird language. We are investigating, for instance: What are the acoustic parameters encoding meaning in these vocalizations? How do finches perceive the meaning of these vocalizations? What are the neuronal computations at stake for extracting the meaning?
The present project aims to identify the perceptual categorization of the vocalizations by zebra finches irrespective of the vocal characteristics of the vocalizers. The idea is to verify the meaning categories that we inferred from observing the birds' behavior by asking the finches themselves. Using an operant conditioning task (a design and protocol already developed in the lab), we want to identify how zebra finches group their vocalizations into semantic categories. We are also going to test the memory capacity of zebra finches.
For Fall 2020, given occupancy restrictions, URAP candidates might not be able to spend much time handling and testing birds. Instead, the URAPs will be trained to perform acoustical analyses of bird communication calls using custom software.

Duties:
Experimental commitment should add up to about 1 hour daily on 5 days out of the week. The apprentice should expect to spend ~1/2 hr to prepare and initiate the tests in the morning (getting conditioning apparatus ready, weighing finches...), and return to spend ~1/2 hour concluding the tests (stopping the tests, returning birds back to the colony...). Additional commitment will involve communicating results and helping to plan further tests. The apprentice will meet regularly with the supervisor to analyze and interpret results, and with the laboratory assistant to discuss bird health and facilities maintenance. Experiments could also be run on the weekends if equipment scheduling allows.
The apprentice will be given a reading list and will be invited to attend weekly lab meetings. At the end of the term, the apprentice will prepare an oral presentation, to be given at one of our lab meetings, about the results of the experiment, as per the URAP contract. Expected outcomes include practicing scientific habits of mind, learning about operant conditioning and social communication behaviors, computational skills performed with guidance in Matlab/Python/R, understanding of statistics, and skills in data representation.

Research Site: Northwest Animal Facility; Research in Fall 2020 will be primarily done off-site., Graduate Student

Qualifications: The undergraduate research apprentice will be involved in both data acquisition and in data analysis. The most important qualification is an interest in animal behavior and vocal communication. Experience with animal behavioral experiments, sound analysis and programming (especially Python) are a plus.

Weekly Hours: 6-8 hrs

Related website: http://theunissen.berkeley.edu/