'

Experiments with Machine Learning - GDG Lviv

Понравилась презентация – покажи это...





Слайд 0

Experiments with Machine Learning Yuriy Guts Solutions Architect


Слайд 1

First Things First What Is Machine Learning?


Слайд 2

“ A computer program is said to learn from experience E with respect to some class of tasks T and performance measure P, if its performance at tasks in T, as measured by P, improves with experience E. — Tom M. Mitchell


Слайд 3

Categories of Machine Learning 1. Supervised Learning. 2. Unsupervised Learning. 3. Reinforcement Learning.


Слайд 4


Слайд 5


Слайд 6

Regression Predict a continuous dependent variable based on independent predictors


Слайд 7


Слайд 8


Слайд 9


Слайд 10

Linear Regression


Слайд 11


Слайд 12


Слайд 13


Слайд 14

Classification Assign an observation to some category from a known discrete list of categories


Слайд 15

Logistic Regression hypothesis  =  1  /  (1  +  exp(-­‐theta'  *  x));


Слайд 16

Logistic Regression: Cost Function hypotheses  =  sigmoid(X  *  theta); cost  =  (1  /  m)  *  (-­‐y'  *  log(hypotheses)  -­‐ (1  -­‐ y)'  *  log(1  -­‐ hypotheses));


Слайд 17

Let’s classify human speech! Decide whether a spoken phrase contains the word ‘Google’ or not


Слайд 18

‘Google’ Detector: Feature Mapping Input: Output: Audio file (WAV, 16 bit mono, 44.1 kHz) 1 if it contains the word ‘Google’, otherwise 0 Options for building X[ ]: 1. Use raw waveform as a feature vector. But: will have 66150 features for a 1.5 second file. Kinda scary, and easy to overfit. 2. Use Mel-Frequency Cepstral Coefficients (MFCC). Believed to be closer to human auditory response. Depending on parameters, can give about 80 features per file.


Слайд 19

[cepstra,   aSpectrum,   pSpectrum]   =  MFCC(waveform); x  =  [cepstra(1);   cepstra(2);   ...;  cepstra(n)];


Слайд 20

Let’s code it up MATLAB, logistic regression with conjugate gradient optimization


Слайд 21

Q&A yuriy . guts  @ gmail . com linkedin . com / in / yuriyguts github.com/YuriyGuts/gdg-speech-classifier


Слайд 22


×

HTML:





Ссылка: