Jonathan Cohen Director, CUDA Libraries and Software Solutions 2 3 - - PowerPoint PPT Presentation

▶
jonathan cohen director cuda libraries and software
SMART_READER_LITE
LIVE PREVIEW

Jonathan Cohen Director, CUDA Libraries and Software Solutions 2 3 - - PowerPoint PPT Presentation

Mar 27, 2024 171 likes •343 views

MACHINE LEARNING: WHAT COMPUTATIONAL RESEARCHERS NEED TO KNOW Jonathan Cohen Director, CUDA Libraries and Software Solutions 2 3 COMPUTERS ARE LEARNING TO SEE! GPU Entries 120 Image Recognition Challenge 100 110 80 1.2M training images

slide-1
SLIDE 1

Jonathan Cohen Director, CUDA Libraries and Software Solutions

MACHINE LEARNING: WHAT COMPUTATIONAL RESEARCHERS NEED TO KNOW

slide-2
SLIDE 2

2

slide-3
SLIDE 3

3

slide-4
SLIDE 4

4

COMPUTERS ARE LEARNING TO SEE!

1.2M training images • 1000 object categories

Hosted by

Image Recognition Challenge

person car helmet motorcycle bird frog person dog chair person hammer flower pot power drill

4 60 110 20 40 60 80 100 120 2010 2011 2012 2013 2014

GPU Entries Classification Error Rates

28% 26% 16% 12% 7% 0% 5% 10% 15% 20% 25% 30% 2010 2011 2012 2013 2014

slide-5
SLIDE 5

5

DEEP NEURAL NETS: ANALYSIS VIA ABSTRACTION

Image “Sara”

slide-6
SLIDE 6

6

Tree Cat Dog Machine Learning Software “turtle” Forward Propagation Compute weight update to nudge from “turtle” towards “dog” Backward Propagation Trained Model “cat” Repeat

Training Classification

slide-7
SLIDE 7

7

Deep Learning for Science

slide-8
SLIDE 8

8

CORAL REEF MAPPING

Coral reefs tremendously important

Support more species per area than any

  • ther marine environment

Storehouse of immense biodiversity Buffer adjacent shorelines from wave action

Ecologists need accurate large-scale coverage, broken down by genus Surveys generate huge data sets…

Material courtesy Oscar Beijbom and CoralNet advisory team / UCSD

slide-9
SLIDE 9

9

LARGE SCALE CORAL ANNOTATION

…but labeling is tedious and slow Anecdotally only 1-2% of image data

  • btained from coral reef surveys is labeled

Automated methods: best methods today work on 60% of data with 5% loss of accuracy Deep learning: estimated to work on 90% of data (pilot study underway) More at: http://coralnet.ucsd.edu/

CoralNet @ UCSD

Material courtesy Oscar Beijbom and CoralNet advisory team / UCSD

Circles are coral genera Acropora,, Pavona,, Montipora,, Pocillopora,,and Porites Triangles are non-coral substrates, Crustose Coralline Algae,,Turf algae,,Macroalgae, and Sand.

slide-10
SLIDE 10

10

CONNECTOME PROJECT

“Imaging neural circuits at nanometer length scales leads inevitably to

vast data sets. In fact, the entire Connectome Project is feasible only now because of the exponential increase in computing power and data storage over time. Nevertheless, managing, visualizing, and analyzing these data remain major challenges.” – Harvard Center for Brain Science website

Mapping the Brain’s Wiring Diagram

slide-11
SLIDE 11

11

Images courtesy Thouis Jones

Microscope generates 0.85 TB / day 250 NVIDIA Tesla K40 GPUs running classifier Next-gen will generate 1GB/sec, running 50% of the time That will be 42TB / day

NEURAL NET SEGMENTING NEURON DATA

Microscopy Image Neural Net Classifier Segmentation

slide-12
SLIDE 12

12

http://www.youtube.com/watch?v=-wq2WTRmeW4 Credits: Daniel Berger and Sebastian Seung (MIT); Narayanan Kasthuri, Richard Schalek, Kenneth Hayworth, Juan-Carlos Tapia and Jeff Lichtman (Harvard).

slide-13
SLIDE 13

13

CANCER CELL MITOSIS DETECTION

Mitosis: Chromosomes in cell nucleus replicated prior to cell splitting “B&R” grading system includes mitotic activity per tissue area – strong indicator of invasive breast cancer Hand count mitosis events over 2mm2 region in stained slide

slide-14
SLIDE 14

14

DEEP NEURAL NETS FOR MITOSIS DETECTION

Use DNNs as pixel classifier input: window of raw pixels

  • utput: probability center pixel is close to

mitosis centroid 2012 contest: 66k pos examples, 6M neg examples 5 months to train on CPU 3 days to train on GPU! See: Dan Ciresan, Mitosis Detection in Breast Cancer Histology Images with Deep Neural Networks – NVIDIA GPU Theater, Tuesday 2:30-2:50

IDSIA – Winner 2012 & 2013 Contests

Material courtesy Dan Ciresan, IDSIA 0.1 0.2 0.3 0.4 0.5 0.6 0.7 F1 score IDSIA

  • ther entries

MICCAI 2013

slide-15
SLIDE 15

15

How to Get Started

slide-16
SLIDE 16

16

ANYONE CAN USE DEEP LEARNING

Several open source frameworks available with active communities Caffe (UCB), Torch (NYU), Theano (Montreal) – take your pick

Caffe: http://caffe.berkeleyvision.org/tutorial/ Torch: http://code.cogbits.com/wiki/doku.php Theano: http://deeplearning.net/software/theano/tutorial/

All have excellent support for NVIDIA GPUs Astronomy, sociology, political science, marine ecology, medical imaging, genomics, plant biology, archaeology, …

Machine Learning Will Impact all of Science

slide-17
SLIDE 17

17

High performance routines for Convolutional Neural Networks Optimized for current and future NVIDIA GPUs Integrated in major open-source frameworks

Caffe, Torch7, Theano

Flexible and easy-to-use API Also available for ARM / Jetson TK1 https://developer.nvidia.com/cuDNN

GPU-ACCELERATED DEEP LEARNING

Caffe (CPU*) 1x Caffe (GPU) 11x Caffe (cuDNN) 14x Baseline Caffe compared to Caffe accelerated by cuDNN on K40

*CPU is 24 core E5-2697v2 @ 2.4GHz Intel MKL 11.1.3