fundamentals of deep learningnvidia, deep learning institute fundamentals of deep learning...
TRANSCRIPT
1
R. Ptucha ‘19 1
Raymond Ptucha, Rochester Institute of TechnologyNVIDIA, Deep Learning Institute
Fundamentals of Deep Learning
Electronic Imaging 2019: SC20January 15, 2019, 8:30am-12:45pm
R. Ptucha ‘19 2
Fair Use AgreementThis agreement covers the use of all slides in this document, please read carefully.
• You may freely use these slides, if:– You send me an email telling me the conference/venue/company name
in advance, and which slides you wish to use.– You receive a positive confirmation email back from me.– My name (R. Ptucha) appears on each slide you use.
(c) Raymond Ptucha, [email protected]
2
R. Ptucha ‘19 4
Agenda
• Part I- Intuition and Theory– 8:35-9:15pm: Introduction– 9:15-10:00pm: Convolutional Neural Networks– 10:00-10:40pm: Recurrent Neural Networks
• 10:40-11:00pm: Break
• Part II- Hands on– 11:00am-12:45pm: Hands-on exercises
DEEP LEARNING FUNDAMENTALS FOR COMPUTER VISION
3
R. Ptucha ‘19 6
• Train first network• Introduce Image
Classification• Necessary ingredients for
successful training
Training
• Role of Neural Networks in Applications
• Building around networks
Deployment• Improving accuracy• Improving capability• Solving novel problems
• Changing layers of network• Beyond Image
Classification
Performance
This Course: Deep Learning Fundamentals
R. Ptucha ‘19 7
Work through Introduction Section
• Navigate to:• courses.nvidia.com/dli-event• Browser Recommendation: Chrome
• Event code: EIC_CV2.0_ AMBASSADOR_JA19• Create an Account• Work through the Introduction Section and
‘Start’ launching your first GPU task
4
R. Ptucha ‘19 99
R. Ptucha ‘19 1111
5
R. Ptucha ‘19 1313
R. Ptucha ‘19 1414
When done viewing slides, scroll down
6
R. Ptucha ‘19 1515
R. Ptucha ‘19 1616
Answer question, then choose next
7
R. Ptucha ‘19 1717
When done viewing slides, scroll down
R. Ptucha ‘19 1818
Click Start
8
R. Ptucha ‘19 1919
It will take a few minutesWhile it is loading, read intro material…
R. Ptucha ‘19 2020
Click Launch Task when ready
9
R. Ptucha ‘19 2121
This will open up a new tab.You can go back and forth between original and this new tab
new taboriginal tab
R. Ptucha ‘19 2222
When you click Open digits, a third tab will open, now you can navigate between all three tabs
10
R. Ptucha ‘19 2323
After the digits window opens, follow directions from this window and perform tasks in digits window.
R. Ptucha ‘19 2424
If you get “lost” in DIGITS, click this to return to this page
Use this to create or work on datasets
Use this to create or work on neural network models
Use this to do transfer learning
11
R. Ptucha ‘19 2525
When you get to the end of the instructions, you are done with that module. You can then close the digitis tab and this module tabDo NOT close the original Deep Neural Networks tabReturn to the original Deep Neural Networks tab
R. Ptucha ‘19 2626
When you get to the original Deep Neural Networks tab, click Stop task, and continue to end of page,.
12
R. Ptucha ‘19 2727
When you get to the end of page, click Next and continue to the next module
R. Ptucha ‘19 28
For Each Module• Step through the slides• Scroll down• Click “start” to launch a GPU session• When ready, click “launch”• This opens a new window for a new module…you go
back and forth between tabs• This new module will have you open DIGITS, creating a
third window…you can go back and forth between all three tabs
• When done with module, hit “stop task”• And continue to bottom of page• Close module and Digits windows, then click Next…
13
R. Ptucha ‘19 29
For More Information:Raymond W. PtuchaAssistant Professor, Computer EngineeringDirector, Machine Intelligence LaboratoryRochester Institute of Technology
Email: [email protected]: +1 (585) 797-5561Office: GLE (09) 3441
http://www.rit.edu/mil