Planning Classification Image Experiments
Planning support using
classification image estimation results from JOV 1(2) papers (Ahumada, 2002; others,
2002).
Outline
Example classification
image experiment: Ahumada, 1996 ECVP
JOV 1(2) Results
Planning Decisions
(iterative process)
Stimulus
representation
Dimensionality
of the representation. m
External
noise distribution and level. s E
Internal
noise level estimation. s I
Number
of trials. N s
Confidence
interval size or power of tests.
We will look at some
alternatives and examples for each decision.
Introduction
Example classification image experiment:
Ahumada, 1996 ECVP.
JOV 1(2) and Newer* Results
(Assuming linear decision function)
Improved estimates of classification image when there is
response bias. (Murray, Bennett, and Sekuler).
Improved estimates of errors in classification image using
estimates of internal noise (Murray, Bennett,
and Sekuler).
Improved methods for specific hypothesis testing (Abbey and Eckstein; Solomon).
Methods for estimating internal noise from the
signal-to-noise ratio of the classification image (Nykamp
and Ringach).
*Simpler
formulas for estimating internal noise from repeated noise trials experiments.
*Simpler formulas for estimating internal noise from the
signal-to-noise ratio of the classification image.
Planning Decisions (iterative
process)
Stimulus representation
where the classification is expected to have a large linear component.
Example
problem: Low vs. high spatial frequency;
Example
solutions:
Auditory frequency
component amplitude;
Solomon FFT power;
Contrast energy (Neri and Heeger, 2002 Nature Neuroscience).
General
nonlinear search spaces:
E. Sutter, Wiener kernel analysis;
S. Klein, Book chapter.
Dimensionality of the
representation. m
External noise
distribution and level. s E
Noise correlation matrix: white, pink, narrow-band,
low-pass
Dynamic range considerations
Masking considerations:
Gain control (internal noise) from linearly irrelevant components.
Mimicking internal noise can mean more “ecological”
validity.
Distribution: spectrum (Ahumada and Watson, 1985; Solomon, 2000) ;
Level: Beard and Ahumada (1997; 1998)
criterion: “no change in performance”.
Internal noise level
estimation. s I
Frozen noise method
(Burgess and Colborne, 1988);
Number of trials (per
stimulus). N S.
Size of classification
image pixel confidence intervals or the power for correctly selecting among
alternative hypotheses.
Multiply per trial signal-to-noise ratio in
classification image by N S
(The
dimensionality of the representation m comes through by reducing the size of
the normalized weights)
The
number of parameters being estimated controls the necessary sample size, not
the number of pixels.
Testing specific hypotheses (Abbey
and Eckstein; Solomon).