1. https://www.mathworks.com/matlabcentral/fileexchange/6626-passive-mode-ftp-in-matlab?focused=5174243&tab=function에서 프로그램을 다운받는다

2. matlab prompt에 'which ftp'라고 입력하여 built-in ftp함수가 있는 폴더를 알아낸다.

나의 경우엔,

C:\Program Files\MATLAB\R2016a\toolbox\matlab\iofun\@ftp\ftp.m  % ftp constructor

3. C:\Program Files\MATLAB\R2016a\toolbox\matlab\iofun 로 이동하여 @ftp라는 폴더를 복사하여 임의의 폴더에 저장한다. e.g., C:\myWork\myFunctions\@ftp

4. matlab prompt에 아래와 같이 입력하여 path를 저장한다

       addpath('C:\myWork\myFunctions')
       savepath

5. PassiveFTP/connect.m 을 'C:\myWork\myFunctions/@ftp/private/'에 복사한다

6. PassiveFTP/active.m dataMode.m  ftp.m  pasv.m 을 'C:\myWork\myFunctions/@ftp/에 복사한다.

7. matlab을 종료했다가 다시 실행한 후, rehash toolboxcache을 입력한다.

8.

myFTP = ftp('ftp.mathworks.com')
    pasv(myFTP)
    dataMode(myFTP)  % this command simply shows the current mode

    you can return to normal/active mode by issuing the command
    active(myFTP)

Some Matlab Code

This collection of Matlab code is brought to you by the phrases "caveat emptor" and "quid quid latine dictum sit, altum videtur", and by the number 404.

Wrapper Code for LIBSVM

batchtest.m : Runs batches of train+test tasks using LIBSVM (Chang & Lin 2000), including model selection for the RBF kernel. Uses svmlwrite.m (by Anton Schwaighofer) and getcm.m. If you ask for class probability estimates, you will also need readprobest.m.

nfolds.m : Combines makefolds.m and batchtest.m

readprobest.m : Reads class probability estimates output when you use the "-b" option with LIBSVM.

splitbysomething.m : Like batchtest, but allows you to create different classifiers for different partitions of all the data. It requires the function ispartition.m.

General Machine Learning

makefolds.m : creates indices of train and test sets for each of n folds for deterministic cross-validation.

makesplits.m : partitions a vector by its values. makefolds could be made shorter if it made calls to makesplits, but I'm not going to change that (now).

getcm.m : Gets confusion matrices, accuracy, precision, recall, F-score, from actual and predicted labels.

precrec.m : Produces precision-recall and ROC curves given true labels and real-valued classifier output. Only for binary classifiers. Uses getcm.

choosebestscore.m : Chooses highest score (given precomputed scores) of elements in sequences.

revhmm.m : Does Viterbi decoding given posteriors.

Manifold Stuff

KADRE.zip : Kick-ass dimensionality reduction (versions of Laplacian Eigenmaps and Isomap that use the Cover Tree algorithm to create the nearest neighbor graph... covertree code only works on linux)

Speech Stuff

makepitchtier.m : creates PitchTier file (for Praat) given measurement times and values.

Visualization / Partiview

Ndaona : package for producing partiview files for interactive 3d or 3d+time models

partiviewangles.m Calculates coordinates Rx and Ry required by Partiview to position a camera. (Very useful.... but use tfm.pl instead.)

Sequence Data

For a description of the SEQ (and FLAT) formats that I use to store data for sequential classification, see my Sequential Data / Classification page.

seqread.m : reads from a file with SEQ data to a SEQ Matlab structure.

seqread.m : writes to a file with SEQ data from a SEQ Matlab structure.

seq2flat : converts from a SEQ Matlab structure to a FLAT Matlab structure.

seqkeepcols.m : converts from a SEQ Matlab structure to another SEQ Matlab structure but with only selected columns.

Tone Recognition

For a description of the PWS (Phrase-Word-Syllable) format that I use to store data for tone recognition, see my Tone Recognition page.

pws2textgrid.m : creates Praat-formatted TextGrid files, with three tiers for Phrase, Word, and Syllable, of a PWS structure (or a cell array thereof).

textgrid2pws.m : reads Praat-formatted TextGrid files into a PWS structure. Good for replacing automatic alignments with manually corrected alignments. (Remember to press SHIFT while doing the manual alignment so that boundaries across different tiers remain aligned!)