-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy pathparams_GradDesc.m
More file actions
27 lines (22 loc) · 1.17 KB
/
Copy pathparams_GradDesc.m
File metadata and controls
27 lines (22 loc) · 1.17 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
function param = params_GradDesc()
% This function sets parameters for the Gradient descent experiment. To
% be used with the script GradDesc_AURORA. PBS 2-14-14
% Data file selection
%param.trainDir = '../data/samples_aurora/data_aurora_train/';
%param.trainFiles = [1:5,56:60]; % male files are 56:109
param.trainDir = '../data/samples_aurora/data_aurora_test/';
param.trainFiles = [1:5,53:57]; % male files are 53:103
param.testDir = '../data/samples_aurora/data_aurora_test/';
param.testFiles = [1:5,53:57]; % male files are 53:103
% Noise parameters for testing and training data
param.noiseTypes = 1:4; % 4 types total
param.noiseLevs = 5; % -5:5:20 dB available
% Peripheral model options
param.periph = 'gamma'; % spec-spectrogram gamma-gammatone meddis-meddis model NSL-NSL spectrogram
param.int = 10; % resampling interval in ms
param.delay = 10; % number of time delay points
param.normType = 'norm'; % remove mean and normalize to unit variance
% Pre-processing options
param.nDim = 50; % dimensions kept in dimension reduction (defaults to all)
param.whiten = 1; % sets whether to whiten the data
end