2015 MATLAB & SIMULINK · 3 Signal Processing and Analysis of Sensor Data Do you have access to...

10
2015 MATLAB & SIMULINK

Transcript of 2015 MATLAB & SIMULINK · 3 Signal Processing and Analysis of Sensor Data Do you have access to...

Page 1: 2015 MATLAB & SIMULINK · 3 Signal Processing and Analysis of Sensor Data Do you have access to sensor data and you need to extract features for building machine learning algorithms?

2015 MATLAB & SIMULINK

Page 2: 2015 MATLAB & SIMULINK · 3 Signal Processing and Analysis of Sensor Data Do you have access to sensor data and you need to extract features for building machine learning algorithms?

2© 2015 The MathWorks, Inc.

Introduction to Signal Processing for

Sensor Analytics

Giorgia Zucchelli – Technical Marketing RF & Mixed-Signal

[email protected]

Page 3: 2015 MATLAB & SIMULINK · 3 Signal Processing and Analysis of Sensor Data Do you have access to sensor data and you need to extract features for building machine learning algorithms?

3

Signal Processing and Analysis of Sensor Data

Do you have access to sensor data and you need to

extract features for building machine learning algorithms?

Is your problem open-ended and you need to explore

alternative solutions?

You are not an expert in signal processing, nor a

professional programmer but you have an idea for a killer

App?

Page 4: 2015 MATLAB & SIMULINK · 3 Signal Processing and Analysis of Sensor Data Do you have access to sensor data and you need to extract features for building machine learning algorithms?

4

Signal Analysis for ClassificationApplication examples

Mobile sensing

Structural health monitoring (SHM)

Automated trading

Engine event detection

Radar post-processing

Advanced surveillance

...

Page 5: 2015 MATLAB & SIMULINK · 3 Signal Processing and Analysis of Sensor Data Do you have access to sensor data and you need to extract features for building machine learning algorithms?

5

What We Will Learn Today

Basic signal manipulation

Statistical parameters estimation

Digital filters

Frequency-domain transformations

Measurements automation and feature extraction

Page 6: 2015 MATLAB & SIMULINK · 3 Signal Processing and Analysis of Sensor Data Do you have access to sensor data and you need to extract features for building machine learning algorithms?

6

Human Activity Analysis and Classification

ClassificationFeature

Extraction

Dataset courtesy of:

Davide Anguita, Alessandro Ghio, Luca Oneto, Xavier Parra and Jorge L. Reyes-Ortiz.

Human Activity Recognition on Smartphones using a Multiclass Hardware-Friendly Support Vector Machine.

International Workshop of Ambient Assisted Living (IWAAL 2012). Vitoria-Gasteiz, Spain. Dec 2012

http://archive.ics.uci.edu/ml/datasets/Human+Activity+Recognition+Using+Smartphones

Page 7: 2015 MATLAB & SIMULINK · 3 Signal Processing and Analysis of Sensor Data Do you have access to sensor data and you need to extract features for building machine learning algorithms?

7

Human Activity Analysis and ClassificationSolution

Only core built-in Signal Processing algorithms

66 high-quality features extracted with only

65 lines of code!

Visualisation and automation

accelerate insight iterations

Page 8: 2015 MATLAB & SIMULINK · 3 Signal Processing and Analysis of Sensor Data Do you have access to sensor data and you need to extract features for building machine learning algorithms?

8

What We Learned:

Use Signal Processing Toolbox functions to extract features

from your signals without re-inventing the wheel

– Cheby2, filter, rms, pwelch, periodogram, xcov, findpeaks

Build machine learning algorithms with Neural Network

Toolbox and Statistics Toolbox

– Classification

Page 9: 2015 MATLAB & SIMULINK · 3 Signal Processing and Analysis of Sensor Data Do you have access to sensor data and you need to extract features for building machine learning algorithms?

9

Conclusions

Extensive set of de-facto

standard functions for

signal processing and analysis

Visualisation and App-driven

automation

accelerate insight iterations

Compact and concise language,

and

extensive documentation