SVM Rules of Thumb
Description
We provide some intuition into how to use SVMs. We give advice about
choosing the regularization parameter C, the kernel and its parameters, and
the size of the subproblem to solve at each step. Solving machine learning
pro
9.09J/7.29J - Cellular Neurobiology, Spring 2005
Massachusetts Institute of Technology
Department of Brain and Cognitive Sciences
Department of Biology
Instructors: Professors William Quinn and Troy Littleton
Problem Set #3
1. Define the relevance of the
9.09J/7.29J - Cellular Neurobiology, Spring 2005
Massachusetts Institute of Technology
Department of Brain and Cognitive Sciences
Department of Biology
Instructors: Professors William Quinn and Troy Littleton
9.09J/7.29J - Cellular Neurobiology, Spring 2005
Massachusetts Institute of Technology
Department of Brain and Cognitive Sciences
Department of Biology
Instructors: Professors William Quinn and Troy Littleton
A systems neuroscience
approach to memory
Critical brain structures for declarative memory
Relational memory vs. item memory
Recollection vs. familiarity
Recall vs. recognition
What about PDs?
R-K paradigm
Associative learning
Why are lesion studies impo
Paper Four: Revision
DUE TO YOUR TA BY 9AM ON FRI, DEC 3
If you write a newspaper story, a grant proposal, or a piece of prose for a textbook, it is very
likely that the first draft will not be the last. Your deathless prose will go to an editor/reviewer
Paper Three: Rewriting the textbook
DUE TO YOUR TA BY 9AM ON FRI, NOV 5
Paper Three must be drawn from the materials relating to Lectures 17-24 / Chapters 13, 15-17
For this paper, you goal is to draft a new piece for the Gleitman et al. textbook. The cen
What is Memory?
Memory properis the knowledge of a former state of mind after it has already once
dropped from consciousness; or rather it is the knowledge of an event, or fact, of which
meantime we have not been thinking, with the additional consciousnes
Paper Two: Taking the next step
DUE TO YOUR TA BY 9AM ON FRI, OCT 8
Paper Two must be drawn from the materials relating to Lectures 7-16 / Chapters 8-12, & 14
For this paper, you goal is to identify a recent finding in the literature, come up with, and
pr
Lecture 18: Thomas Serre
Supplementary reading list
T. Serre. Learning a dictionary of shape-components in visual cortex:
Comparison with neurons, humans and machines, PhD Thesis, May 2006
T. Serre, L. Wolf and T. Poggio. Object recognition with features
Paper One: Writing for the public
DUE TO YOUR TA BY 9AM ON FRI, SEPT 24
Topics: Note: You can write about a topic before you have heard the lecture on the topic
The Paper One topic must be drawn from the materials relating to Lectures 1-6 / Chapters 1-6
F
Lecture 09:
Sayan Mukherjee
Description
Necessary and sufficient requirements for uniform convergence for both real-valued loss
functions and classification are introduced. VC entropy, VC dimension, empirical
covering numbers, and V-gamma dimension are in
MIT OpenCourseWare
http:/ocw.mit.edu
24.08J / 9.48J Philosophical Issues in Brain Science
Spring 2009
For information about citing these materials or our Terms of Use, visit: http:/ocw.mit.edu/terms.
9.48/24.08J S09!
Handout
Arguments: the basics
An argum
Your schedule of coming weeks
Today: One-way ANOVA, part II
Next week:
Two-way ANOVA, parts I and II.
One-way ANOVA HW due Thursday
Week of May 4
Teacher out of town all week
No class on Tuesday, 5/4
Thursday class: TAs talk about statistical learning
Lecture 16:
Alex Rakhlin
Description
We relate consistency or empirical risk minimization (ERM) with uniform convergence
in probability over function classes. We prove generalization bounds for ERM in a
bounded RKHS.
Suggested Reading
V. N. Vapnik. The Na
MIT OpenCourseWare
http:/ocw.mit.edu
24.08J / 9.48J Philosophical Issues in Brain Science
Spring 2009
For information about citing these materials or our Terms of Use, visit: http:/ocw.mit.edu/terms.
24.200: Ancient Philosophy
Prof. Sally Haslanger
Meno a
Review
One-way ANOVA, I
9.07
4/15/2004
Multiple comparisons
We often need a tool for comparing more
than two sample means
Earlier in this class, we talked about twosample z- and t-tests for the difference
between two conditions of an independent
variabl
Lecture 15: Stability of Tikhonov
Regularization
Alex Rakhlin
Description
We briefly review the generalization bounds of last lecture before turning to our main
goal - using the stability approach to prove generalization bounds for Tikhonov
regularization
MIT OpenCourseWare
http:/ocw.mit.edu
24.08J / 9.48J Philosophical Issues in Brain Science
Spring 2009
For information about citing these materials or our Terms of Use, visit: http:/ocw.mit.edu/terms.
GUIDELINES FOR PAPERS
1. All papers should be neatly ty
Summary sheet from last time:
Confidence intervals
Confidence intervals take on the usual form: parameter =
statistic tcrit SE(statistic)
parameter
a
b
SE
se sqrt(1/N + mx2/ssxx)
se/sqrt(ssxx)
y (mean)
se sqrt(1/N + (xo mx)2/ssxx)
ynew
(individual)
se s
Lecture 14: Generalization Bounds and
Stability
Alex Rakhlin
Description
We introduce the notion of generalization bounds, which allow us to have confidence in
the functions our algorithms are finding. We introduce the notion of algorithmic stability,
and
Outline Correlation & Regression, III
9.07 4/6/2004 Relationship between correlation and regression, along with notes on the correlation coefficient Effect size, and the meaning of r Other kinds of correlation coefficients Confidence intervals on
MIT OpenCourseWare
http:/ocw.mit.edu
24.08J / 9.48J Philosophical Issues in Brain Science
Spring 2009
For information about citing these materials or our Terms of Use, visit: http:/ocw.mit.edu/terms.
Third paper. NB: no paraphrases; answer in your own wor
MIT OpenCourseWare
http:/ocw.mit.edu
24.941J / 6.543J / 9.587J / HST.727J The Lexicon and Its Features
Spring 2007
For information about citing these materials or our Terms of Use, visit: http:/ocw.mit.edu/terms.
Categorical perception and its
implication
Steps in regression analysis (so far)
Correlation & Regression, II
9.07
4/6/2004
Residual Plots
Plotting the residuals (yi yi) against xi can
reveal how well the linear equation explains
the data
Can suggest that the relationship is
significantly non-li
MIT OpenCourseWare
http:/ocw.mit.edu
24.08J / 9.48J Philosophical Issues in Brain Science
Spring 2009
For information about citing these materials or our Terms of Use, visit: http:/ocw.mit.edu/terms.
Second paper. NB: no paraphrases; answer in your own wo
Regression and correlation
Involve bivariate, paired data, X & Y
Correlation & Regression, I
9.07
4/1/2004
Regression & correlation
Concerned with the questions:
Does a statistical relationship exist between X & Y,
which allows some predictability of o
Lecture 12: Online Learning
Andrea Caponnetto, Sanmay Das
Description
To introduce the general setting of online learning. To describe an online version of the
RLS algorithm and analyze its performance. To discuss convergence results of the
classical Perc
MIT OpenCourseWare
http:/ocw.mit.edu
24.08J / 9.48J Philosophical Issues in Brain Science
Spring 2009
For information about citing these materials or our Terms of Use, visit: http:/ocw.mit.edu/terms.
First paper. NB: no paraphrases; answer in your own wor