Code

Accuracy Maximization Analysis (AMA) - Matlab and C code implementation

Description: Computes the posterior probability of a category given the filter response(s) to the stimulus

•Geisler WS, Najemnik J, Ing AD (2009). Optimal stimulus encoders for natural tasks. Journal of Vision, 9(13):17, 1-16 [ pdf | correction ]

•Burge J, Geisler WS (2011). Optimal defocus estimation in individual natural images. Proceedings of the National Academy of Sciences, 108 (40): 16849-16854 [ pdf ]

Relevant publications: Please cite the following papers if the code on this page contributes to your research

Logic:

(4) Response Distributions of All Stims in All Ctgs: N(r(i,j),sigma(i,j))

(7) Posterior Probability of Correct Category given s(k,l)

(3) Response Distributions of All Stims in Same Ctg: N(r(k,j),sigma(k,j))

(6) Likelihood of Mean Response r(k,l) from s(k,j) and s(i,j)

(5) Likelihood of Mean Response r(k,l) from s(k,l)

(2) Response Distribution of that Stimulus: N(r(k,l),sigma(k,l))

(1) Select stimulus: s(k,l)

Use: AMA requires computing the posterior probability of each stimulus category given the filter responses

for every stimulus in the training set

To use AMA, a minimization routine and an objective function must be chosen

Note that the minimization routine and objective function are not fundamental to the AMA procedure

To get you started, an example minimization routine and an example objective function are provided

However, other minimization routines and/or objective functions may better suit your needs

The example below uses Matlab’s fmincon.m to find AMA filters that minimize the average relative

entropy of the posterior probability distributions over the categories, across all stimuli in the training set

%%% OBJECTIVE FUNCTION %%%

function E = objFunc(w,s,ctgInd)

% function E = objFunc(s,w,ctgInd)

% objective function for AMA

% s: stimuli. vector magnitude of each stimulus must equal 1 [numStims x numDimensions]

% w: filters. vector magnitude of each filter must equal 1 [numDimensions x numFilters]

% ctgInd: category membership of each stimulus [numDimensions x 1]

% %%%%%%

% E: objective to minimize

r = stim2resp(s,w,5.7); % mean response from filter weights and stimuli

sigma = resp2sigma(r,1.36,0.23); % sigma from mean response

pp = AMAengine(r,sigma,ctgInd); % posterior probability of correct category

E = -sum((1/size(s,1)).*log2(pp)); % average relative entropy in bits

(1) Download and unzip AMAengine.zip
(2) Put all five unzipped files in the current folder and/or in a folder on the Matlab path
(3) At the command prompt type: >> mex AMAengine.cpp
(4) At the command prompt type: >> help AMAengine
(5) At the command prompt type: >> AMAengineTest;