9.40 Intro to Neural Computation
Spring 2015
Instructors: Michale S Fee, Daniel Zysman
TAs: Gregory A. Ciccarelli, Benjamin Matthew Deen, Alexander J E Kell, Emily Lambert Mackevicius, Eduardo A Maury, Abra H Shen
Lecture: TR10.30-12 (46-3002)
Announcements
Summer tutorials in computational neuroscience topics
Dear class,This summer, I've organized a series of tutorials related to BCS topics. The tutorials are geared towards grad students and postdocs, but this class will have prepared you well, so you're definitely welcome to come if you're around. You can sign up here:
https://docs.google.com/forms/d/1MZ9xl6zSA6RgjVbOWfUuGisZgFzBY9obEudEOdTaaP4/viewform
Here is the information:
June 9th, 46-3189, 4pm-6pm: Bayesian methods, brain and cog
perspectives
Taught by: Mehrdad Jazayeri, Josh Tenenbaum, Sam Gershman, Evan
Remington, and Devika Narain
June 10th, 46-3189, 4pm-6pm: Learning in recurrent neural
networks
Taught by: Larry Abbott
June 17th, 46-3189, 4pm-6pm: Learning in deep neural
networks
Taught by: Phillip Isola
Each tutorial will consist of a short lecture, and then 'office hours' time to work through practice problems, and discuss problems people want help with in their own research. Food will be provided.
Also, see some of you at office hours tonight! (7pm, 46-5056)
Emily M.
Announced on 19 May 2015 9:56 a.m. by Emily Lambert Mackevicius
Office hours
Come to 46-4062 for office hours tonight at 7:30.Announced on 23 February 2015 7:16 p.m. by Emily Lambert Mackevicius
Recitation 3
Recitation 3 is today! We will cover:1. Integrate and fire, Activation curve, refractory
period.
2. I-V curves, the role of voltage dependent conductances, basics
of HH.
Review lecture notes and readings. We will be doing problems on the board this time (no MATLAB).
Announced on 18 February 2015 12:28 p.m. by Emily Lambert Mackevicius
Recitation 2 Instructions
Recitation 2 is tomorrow afternoon, and I've posted instructions under Materials>Recitation 2>Recitation 2 instructions.Please read the instructions ahead of time, and remember to bring your laptop. We will be using MATLAB again.
If you have questions before recitation, please post on piazza.
Announced on 10 February 2015 2:07 p.m. by Emily Lambert Mackevicius
Rescheduled: Emily's brain lunch talk
My talk has been rescheduled to next Tuesday, February 17th @12pm in 46-3189. Lunch is provided. Here's the abstract:Brain Lunch
Speaker: Emily Mackevicius, Fee Lab
TITLE: Aligning auditory and motor representations of syllable onsets in songbird vocal learning
ABSTRACT: How does the brain learn to produce complex sequences
of behavior? Does it break them down into simpler pieces? How are
behavioral goals represented, and translated into the actions that
produce those goals? Songbirds learn to imitate the songs of tutors
they heard as juveniles. Juvenile song starts off as unstructured
babbling, then becomes more structured with the introduction of a
stereotyped ‘protosyllable’. New syllables emerge through the
differentiation of this protosyllable into multiple syllable types,
until the song crystalizes into an adult song composed of 3-7
distinct syllables.
My work focuses on understanding how the brain of a young bird
learns the motor commands for each new syllable it acquires. My
recordings suggest that auditory and motor representations of song
are both organized around syllables as units. Specifically, I have
recorded in nucleus interface (NIf), a cortical region at the
interface between the auditory and motor systems, and find NIf
neurons burst at or just before the onsets of syllables both during
singing and when juveniles are listening to their tutor song. Some
NIf neurons are selective for particular syllable types. By
aligning to syllable onsets, NIf could “chunk” songs into syllables
each of which is accessible as a discrete unit in both auditory and
motor representations. I have analyzed the role of NIf in the
emergence of motor programs for new syllable types in downstream
premotor circuits, and suggest a simple mechanism by which timing
learned in one domain (listening to a tutor) could be transferred
to support learning in a new domain (singing). I will describe both
experimental and modeling results.
Announced on 05 February 2015 1:08 p.m. by Emily Lambert Mackevicius