Upload File

demystifying deep learning...4 deep learning model learns to perform classification tasks directly...

  • Home
  • Documents
  • Demystifying Deep Learning...4 Deep Learning Model learns to perform classification tasks directly from data. x1000 x1000 x1000 x1000 x1000 Deep Learning Model Image Classifier 6 Deep
41
1 © 2015 The MathWorks, Inc. Demystifying Deep Learning Emelie Andersson

Upload: others

Post on 08-Apr-2020

6 views

Category:

Documents


0 download

Report

TRANSCRIPT

Page 1: Demystifying Deep Learning...4 Deep Learning Model learns to perform classification tasks directly from data. x1000 x1000 x1000 x1000 x1000 Deep Learning Model Image Classifier 6 Deep

1© 2015 The MathWorks, Inc.

Demystifying Deep Learning

Emelie Andersson

Page 2: Demystifying Deep Learning...4 Deep Learning Model learns to perform classification tasks directly from data. x1000 x1000 x1000 x1000 x1000 Deep Learning Model Image Classifier 6 Deep

2

Page 3: Demystifying Deep Learning...4 Deep Learning Model learns to perform classification tasks directly from data. x1000 x1000 x1000 x1000 x1000 Deep Learning Model Image Classifier 6 Deep

3

What is Deep Learning?

Page 4: Demystifying Deep Learning...4 Deep Learning Model learns to perform classification tasks directly from data. x1000 x1000 x1000 x1000 x1000 Deep Learning Model Image Classifier 6 Deep

4

Deep Learning

Model learns to perform classification tasks directly from data.

x1000

x1000

x1000

x1000

x1000

Deep

Learning

Model

Image

Classifier

Page 5: Demystifying Deep Learning...4 Deep Learning Model learns to perform classification tasks directly from data. x1000 x1000 x1000 x1000 x1000 Deep Learning Model Image Classifier 6 Deep

5

Data Types for Deep Learning

Signal ImageText

Page 6: Demystifying Deep Learning...4 Deep Learning Model learns to perform classification tasks directly from data. x1000 x1000 x1000 x1000 x1000 Deep Learning Model Image Classifier 6 Deep

6

Deep Learning is Versatile

Iris Recognition – 99.4% accuracy2

Rain Detection and Removal1Detection of cars and road in autonomous driving systems

1. Deep Joint Rain Detection and Removal from a Single Image" Wenhan Yang,

Robby T. Tan, Jiashi Feng, Jiaying Liu, Zongming Guo, and Shuicheng Yan

2. Source: An experimental study of deep convolutional features for iris recognition

Signal Processing in Medicine and Biology Symposium (SPMB), 2016 IEEE

Shervin Minaee ; Amirali Abdolrashidiy ; Yao Wang; An experimental study of

deep convolutional features for iris recognition

Page 7: Demystifying Deep Learning...4 Deep Learning Model learns to perform classification tasks directly from data. x1000 x1000 x1000 x1000 x1000 Deep Learning Model Image Classifier 6 Deep

7

How is deep learning performing so well?

Page 8: Demystifying Deep Learning...4 Deep Learning Model learns to perform classification tasks directly from data. x1000 x1000 x1000 x1000 x1000 Deep Learning Model Image Classifier 6 Deep

8

Deep Learning Uses a Neural Network Architecture

Input

Layer Hidden Layers (n)

Output

Layer

Page 9: Demystifying Deep Learning...4 Deep Learning Model learns to perform classification tasks directly from data. x1000 x1000 x1000 x1000 x1000 Deep Learning Model Image Classifier 6 Deep

9

Thinking about Layers

▪ Layers are like blocks

– Stack them on top of each other

– Replace one block with a

different one

▪ Each hidden layer processes

the information from the

previous layer

Page 10: Demystifying Deep Learning...4 Deep Learning Model learns to perform classification tasks directly from data. x1000 x1000 x1000 x1000 x1000 Deep Learning Model Image Classifier 6 Deep

10

Thinking about Layers

▪ Layers are like blocks

– Stack them on top of each other

– Replace one block with a

different one

▪ Each hidden layer processes

the information from the

previous layer

➢ Layers can be ordered in

different ways

Page 11: Demystifying Deep Learning...4 Deep Learning Model learns to perform classification tasks directly from data. x1000 x1000 x1000 x1000 x1000 Deep Learning Model Image Classifier 6 Deep

11

Deep Learning in 6 Lines of MATLAB Code

Page 12: Demystifying Deep Learning...4 Deep Learning Model learns to perform classification tasks directly from data. x1000 x1000 x1000 x1000 x1000 Deep Learning Model Image Classifier 6 Deep

12

Why MATLAB for Deep Learning?

▪ MATLAB is Productive

▪ MATLAB is Fast

▪ MATLAB Integrates with Open Source

Page 13: Demystifying Deep Learning...4 Deep Learning Model learns to perform classification tasks directly from data. x1000 x1000 x1000 x1000 x1000 Deep Learning Model Image Classifier 6 Deep

13

Why MATLAB for Deep Learning?

▪ MATLAB is Productive

▪ MATLAB is Fast

▪ MATLAB integrates with Open Source

Page 14: Demystifying Deep Learning...4 Deep Learning Model learns to perform classification tasks directly from data. x1000 x1000 x1000 x1000 x1000 Deep Learning Model Image Classifier 6 Deep

14

“I love to label and

preprocess my data”

True False

Page 15: Demystifying Deep Learning...4 Deep Learning Model learns to perform classification tasks directly from data. x1000 x1000 x1000 x1000 x1000 Deep Learning Model Image Classifier 6 Deep

15

Caterpillar Case Study

▪ World’s leading manufacturer of

construction and mining

equipment.

▪ Similarity between these

projects?

– Autonomous haul trucks

– Pedestrian detection

– Equipment classification

– Terrain mapping

Page 16: Demystifying Deep Learning...4 Deep Learning Model learns to perform classification tasks directly from data. x1000 x1000 x1000 x1000 x1000 Deep Learning Model Image Classifier 6 Deep

16

Computer Must Learn from Lots of Data

▪ ALL data must first be labeled to create these autonomous systems.

“We were spending way too much time ground-truthing [the data]”

--Larry Mianzo, Caterpillar

Page 17: Demystifying Deep Learning...4 Deep Learning Model learns to perform classification tasks directly from data. x1000 x1000 x1000 x1000 x1000 Deep Learning Model Image Classifier 6 Deep

17

How Did Caterpillar Do with Our Tools?

▪ Semi-automated labeling process

▪ Used MATLAB for entire development workflow.

– “Because everything is in MATLAB, development time is short”

Page 18: Demystifying Deep Learning...4 Deep Learning Model learns to perform classification tasks directly from data. x1000 x1000 x1000 x1000 x1000 Deep Learning Model Image Classifier 6 Deep

18

Speed up labeling with Image Labeler App

Page 19: Demystifying Deep Learning...4 Deep Learning Model learns to perform classification tasks directly from data. x1000 x1000 x1000 x1000 x1000 Deep Learning Model Image Classifier 6 Deep

19

Page 20: Demystifying Deep Learning...4 Deep Learning Model learns to perform classification tasks directly from data. x1000 x1000 x1000 x1000 x1000 Deep Learning Model Image Classifier 6 Deep

20

Page 21: Demystifying Deep Learning...4 Deep Learning Model learns to perform classification tasks directly from data. x1000 x1000 x1000 x1000 x1000 Deep Learning Model Image Classifier 6 Deep

21

Page 22: Demystifying Deep Learning...4 Deep Learning Model learns to perform classification tasks directly from data. x1000 x1000 x1000 x1000 x1000 Deep Learning Model Image Classifier 6 Deep

22

MATLAB is Productive

▪ Image Labeler App semi-automates labeling workflow

▪ Bootstrapping

– Improve automatic labeling by updating algorithm as you label

more images correctly.

Page 23: Demystifying Deep Learning...4 Deep Learning Model learns to perform classification tasks directly from data. x1000 x1000 x1000 x1000 x1000 Deep Learning Model Image Classifier 6 Deep

23

MATLAB is Fast

Performance

Training Deployment

Page 24: Demystifying Deep Learning...4 Deep Learning Model learns to perform classification tasks directly from data. x1000 x1000 x1000 x1000 x1000 Deep Learning Model Image Classifier 6 Deep

24

What is Training?Feed labeled data into neural network to create working model

Convolution

Neural

Network

Image

Classifier

Model

x1000

x1000

x1000

x1000

x1000

Page 25: Demystifying Deep Learning...4 Deep Learning Model learns to perform classification tasks directly from data. x1000 x1000 x1000 x1000 x1000 Deep Learning Model Image Classifier 6 Deep

25

Speech Recognition Example

Audio signal → Spectrogram → Image Classification algorithm

Time Time

Am

plit

ude

Fre

quency

Page 26: Demystifying Deep Learning...4 Deep Learning Model learns to perform classification tasks directly from data. x1000 x1000 x1000 x1000 x1000 Deep Learning Model Image Classifier 6 Deep

26

Side note: Another Network for Signals - LSTM

▪ LSTM = Long Short Term Memory (Networks)

– Signal, text, time-series data

– Use previous data to predict new information

▪ I live in France. I speak ___________.

c0 C1 Ct

Page 27: Demystifying Deep Learning...4 Deep Learning Model learns to perform classification tasks directly from data. x1000 x1000 x1000 x1000 x1000 Deep Learning Model Image Classifier 6 Deep

27

1. Create Datastore

▪ Datastore creates

reference for data

▪ Do not have to load in

all objects into memory

Page 28: Demystifying Deep Learning...4 Deep Learning Model learns to perform classification tasks directly from data. x1000 x1000 x1000 x1000 x1000 Deep Learning Model Image Classifier 6 Deep

28

2. Compute Speech Spectrograms

Am

plit

ude

Fre

quency

Time

Page 29: Demystifying Deep Learning...4 Deep Learning Model learns to perform classification tasks directly from data. x1000 x1000 x1000 x1000 x1000 Deep Learning Model Image Classifier 6 Deep

29

3. Split datastores

Training Validation Test

70% 15% 15%

• Trains the model

• Computer “learns”

from this data

• Checks accuracy

of model during

training

• Tests model accuracy

• Not used until validation

accuracy is good

Page 30: Demystifying Deep Learning...4 Deep Learning Model learns to perform classification tasks directly from data. x1000 x1000 x1000 x1000 x1000 Deep Learning Model Image Classifier 6 Deep

30

4. Define Architecture and Parameters

Neural Network Architecture

Model Parameters

Page 31: Demystifying Deep Learning...4 Deep Learning Model learns to perform classification tasks directly from data. x1000 x1000 x1000 x1000 x1000 Deep Learning Model Image Classifier 6 Deep

31

5. Train Network

Page 32: Demystifying Deep Learning...4 Deep Learning Model learns to perform classification tasks directly from data. x1000 x1000 x1000 x1000 x1000 Deep Learning Model Image Classifier 6 Deep

32

Deep Learning on CPU, GPU, Multi-GPU and Clusters

Single CPU

Single CPUSingle GPU

HOW TO TARGET?

Single CPU, Multiple GPUs

On-prem server with GPUs

Cloud GPUs(AWS)

Page 33: Demystifying Deep Learning...4 Deep Learning Model learns to perform classification tasks directly from data. x1000 x1000 x1000 x1000 x1000 Deep Learning Model Image Classifier 6 Deep

33

MATLAB is Fast for Deployment

▪ Target a GPU for optimal

performance

▪ NVIDIA GPUs use CUDA

code

▪ We only have MATLAB code.

Can we translate this?

Page 34: Demystifying Deep Learning...4 Deep Learning Model learns to perform classification tasks directly from data. x1000 x1000 x1000 x1000 x1000 Deep Learning Model Image Classifier 6 Deep

34

GPU Coder

▪ Automatically generates CUDA Code from MATLAB Code

– can be used on NVIDIA GPUs

Page 35: Demystifying Deep Learning...4 Deep Learning Model learns to perform classification tasks directly from data. x1000 x1000 x1000 x1000 x1000 Deep Learning Model Image Classifier 6 Deep

35

Prediction Performance: Fast with GPU Coder

Why is GPU Coder so fast?

– Analyzes and optimizes

network architecture

– Invested 15 years in code

generation

AlexNet ResNet-50 VGG-16

TensorFlow

MATLAB

MXNet

GPU Coder

Images/Sec

Page 36: Demystifying Deep Learning...4 Deep Learning Model learns to perform classification tasks directly from data. x1000 x1000 x1000 x1000 x1000 Deep Learning Model Image Classifier 6 Deep

36

Why MATLAB?

▪ MATLAB is Productive

▪ MATLAB is Fast

▪ MATLAB Integrates with Open Source

Page 37: Demystifying Deep Learning...4 Deep Learning Model learns to perform classification tasks directly from data. x1000 x1000 x1000 x1000 x1000 Deep Learning Model Image Classifier 6 Deep

37

Used MATLAB and Open Source Together

1. Deep Joint Rain Detection and Removal from a Single

Image" Wenhan Yang, Robby T. Tan, Jiashi Feng,

Jiaying Liu, Zongming Guo, and Shuicheng Yan

▪ Used Caffe and MATLAB

together

▪ Use our tools where it

makes your workflow

easier!

Page 38: Demystifying Deep Learning...4 Deep Learning Model learns to perform classification tasks directly from data. x1000 x1000 x1000 x1000 x1000 Deep Learning Model Image Classifier 6 Deep

38

MATLAB Integrates with Open Source Frameworks

▪ Access to many pretrained models through add-ons

▪ Users wanted to import latest models

▪ Import models directly from Tensorflow or Caffe

– Allows for improved collaboration

Page 39: Demystifying Deep Learning...4 Deep Learning Model learns to perform classification tasks directly from data. x1000 x1000 x1000 x1000 x1000 Deep Learning Model Image Classifier 6 Deep

39

Keras-Tensorflow Importer

Page 40: Demystifying Deep Learning...4 Deep Learning Model learns to perform classification tasks directly from data. x1000 x1000 x1000 x1000 x1000 Deep Learning Model Image Classifier 6 Deep

40

MATLAB Integrates with Open Source

Frameworks

▪ MATLAB supports entire deep learning workflow

– Use when it is convenient for your workflow

▪ Access to latest models

▪ Improved collaboration with other users

Page 41: Demystifying Deep Learning...4 Deep Learning Model learns to perform classification tasks directly from data. x1000 x1000 x1000 x1000 x1000 Deep Learning Model Image Classifier 6 Deep

41

Why MATLAB for Deep Learning?

▪ MATLAB is Productive

▪ MATLAB is Fast (Performance)

▪ MATLAB Integrates with Open Source

(Frameworks)


THE DEEP LEARNING REVOLUTION - Nvidia...THE DEEP LEARNING REVOLUTION Superhuman Breakthroughs in Modern Artificial Intelligence Powered by GPUs JOIN THE DEEP LEARNING ERA DEEP LEARNING

From Reinforcement Learning to Deep Reinforcement …fagostin/assets/files/...Keywords: Machine learning · Reinforcement learning Deep learning · Deep reinforcement learning 1 Introduction

Deep learning / representation learning

Binary Deep Learning - TAUweb.eng.tau.ac.il/.../wp-content/uploads/2017/03/Binary-Deep-Learning.pdf · Binary Deep Learning Deep Learning Seminar, School of Electrical Engineering,

Deep Learning I - Korea Universitydsba.korea.ac.kr/wp/wp-content/seminar/Deep learning/Deep Learning... · Deep Learning I Deep learning is research on learning models with multilayer

Intel Quark SoC X1000 - Farnell element14 · Document Number: 329676-005US Intel ® Quark SoC X1000 Datasheet August 2015

Executing Deep Learning Strategies Masterclass Preview - Enterprise Deep Learning

Deep Learning Lab Course 2017 (Deep Learning …ais.informatik.uni-freiburg.de/teaching/ws17/deep...Deep Learning Lab Course 2017 (Deep Learning Practical) Labs: (Computer Vision)

Introducing Deep Learning with MATLAB - MathWorks · Introducing Deep Learning with MATLAB7 How A Deep Neural Network ... relevant features of an image. With deep learning, ... Deep

DENON AVR-E300,X1000,X1010.pdf

Deep Learning: An Introduction from the NLP Perspective Deep Learning - An...Deep Learning Approach 1 Deep Learning Approach 2 Disclaimer I am not (yet) an expert in Deep Learning

An Introduction to Deep Learning · Deep Learning An Introduction to ... •Deep Learning •Feed Forward Network •Problems of Deep Learning •AutoEncoders •Restricted Boltzmann

Lecture 1: Introduction to Deep Learning - uvadlc.github.io · Lecture 1: Introduction to Deep Learning Efstratios Gavves. UVA DEEP LEARNING COURSE UVA DEEP LEARNING COURSE –EFSTRATIOS

Learning Deep Learning

Firebox x1000

20151216 Computer Vision Meetup Deep Learning Deep Learning - Rene...Deep Learning . René Donner Deep Learning Overview 2 ... coursera ... 20151216 Computer Vision Meetup Deep Learning

Introduction to Deep Learning - Computer Graphics€¦ · Introduction to Deep Learning CS468 Spring 2017 Charles Qi. What is Deep Learning? Deep learning allows computational models

State 2Year 2 - eKonomics · State 2Year 2 MODEL- PO 5 removal, x1000 tons MODEL-K 2O removal, x1000 tons MODEL-P 2O Consumption, x1000 tons MODEL-K 2O Consumption, x1000 tons MODEL-PO

Deep materials informatics: Applications of deep learning in … · deep learning: † Deep learning requires big data: In many cases, the biggest limiting factor for applying deep

Deep Learning with H2O - University of Idahostevel/504/Deep Learning with...Deep Learning with H2O. . 4Deep Learning Overview Unlike the neural networks of the past, modern Deep Learning

Deep Learning for Reinforcement Learning in · PDF fileDeep Learning for Reinforcement Learning in ... Deep Learning for Reinforcement Learning in Pacman Deep Learning für ... Während

Lecture 1: Introduction to Deep Learning · uva deep learning course –efstratios gavves introduction to deep learning - 1

Deep Learning for Reinforcement Learning in Pacman · Deep Learning for Reinforcement Learning in Pacman Deep Learning für Reinforcement Learning in Pacman Vorgelegte Bachelor-Thesis

Deep Learning Early Work Why Deep Learning Stacked Auto Encoders Deep Belief Networks CS 678 – Deep Learning1

Deep Learning - South Dakota State UniversityDeep Learning Deep learning (also known as deep structured learning, hierarchical learning or deep machine learning) is a branch of machine

Deep Learning for Generic Object Detection: A Survey · deep learning, probabilistic models, autoencoders, manifold learning, and deep networks 17 Deep learning LeCun et al. (2015)

Demystifying Deep Learning - MathWorks · Demystifying Deep Learning. 2 What’s Deep Learning and why should I care? A practical approach to Deep Learning (for images) –Transfer

Intro to Deep Learning - bpTT Bright Minds … · 17/08/2020  · Intro to Deep Learning Deep Sequence Modeling Deep Computer Vision Deep Generative Modeling Deep Learning Applications

UNSUPERVISED DEEP LEARNING - TAUweb.eng.tau.ac.il/.../uploads/2016/12/Unsupervised-Deep-Learning.pdf · UNSUPERVISED DEEP LEARNING Erez Aharonov Noam Eilon Deep Learning Seminar School

Deep Learning for Laser Based Odometry Estimation - …juxi.net/workshop/deep-learning-rss-2016/papers/Nicolai - Deep... · Deep Learning for Laser Based Odometry Estimation ... deep

Hierarchical Deep Reinforcement Learning: Integrating ...papers.nips.cc/paper/6233-hierarchical-deep-reinforcement-learning... · work that integrates deep reinforcement learning

Deep Learning in real world @Deep Learning Tokyo

Tutorial: Deep Reinforcement Learning · Outline Introduction to Deep Learning Introduction to Reinforcement Learning Value-Based Deep RL Policy-Based Deep RL Model-Based Deep RL

R The Specialist in Value-based Fastening GLIN-X1000 & … · 2019. 2. 27. · R 1 The Specialist in GLIN-X1000 Value-based Fastening & Assembly Solutions NGES GLIN-X1000 Heavy Duty

IMA 1000 Watts Power Supply Series...IMA-x1000 Power Supply Series Main DC Output (J2) IMA-x1000-12 IMA-x1000-24 IMA-x1000-48 Nominal output voltage 12 V 24 V 48 VOutput voltage adjustment