Subscribe to DSC Newsletter

Squeezing Deep Learning into Mobile Phones - A Practitioner's guide

This is a slideshare presentation by Anirudh Koul. Anirudh is deep learning data scientist at Microsoft AI & Research. He earned a master of computational data science at Carnegie Mellon University, and a graduate certificate in data mining from Stanford University. He currently lives in the Bay Area. Anirudh is leading projects like Seeing AI (for the blind community) and others.

Yesterday, I gave a talk at the Strata+Hadoop World Conference on “Squeezing Deep Learning into Mobile Phones - A Practitioner's guide”. Luckily, it seems to have organically gone viral on Twitter, with 3000 views in 12 hours. I thought your readers might find it interesting too, hence sharing it with you.

Below is a transcript of the presentation: 

  • 1. Squeezing Deep Learning into mobile phones - A Practitioners guide Anirudh Koul
  • 2. Anirudh Koul , @anirudhkoul , http://koul.ai Project Lead, Seeing AI Applied Researcher, Microsoft AI & Research Akoul at Microsoft dot com Currently working on applying artificial intelligence for productivity, augmented reality and accessibility Along with Eugene Seleznev, Saqib Shaikh, Meher Kasam
  • 3. Why Deep Learning On Mobile? Latency Privacy
  • 4. Mobile Deep Learning Recipe Mobile Inference Engine + Pretrained Model = DL App (Efficient) (Efficient)
  • 5. Building a DL App in _ time
  • 6. Building a DL App in 1 hour
  • 7. Use Cloud APIs Microsoft Cognitive Services Clarifai Google Cloud Vision IBM Watson Services Amazon Rekognition
  • 8. Microsoft Cognitive Services Models won the 2015 ImageNet Large Scale Visual Recognition Challenge Vision, Face, Emotion, Video and 21 other topics
  • 9. Building a DL App in 1 day
  • 10. http://deeplearningkit.org/2015/12/28/deeplearningkit-deep-learning... Energy to train Convolutional Neural Network Energy to use Convolutional Neural Network
  • 11. Base PreTrained Model ImageNet – 1000 Object Categorizer Inception Resnet
  • 12. Running pre-trained models on mobile MXNet Tensorflow CNNDroid DeepLearningKit Caffe Torch
  • 13. MXNET Amalgamation : Pack all the code in a single source file Pro: • Cross Platform (iOS, Android), Easy porting • Usable in any programming language Con: • CPU only, Slow https://github.com/Leliana/WhatsThis
  • 14. Tensorflow Easy pipeline to bring Tensorflow models to mobile Great documentation Optimizations to bring model to mobile Upcoming : XLA (Accelerated Linear Algebra) compiler to optimize for hardware
  • 15. CNNdroid GPU accelerated CNNs for Android Supports Caffe, Torch and Theano models ~30-40x Speedup using mobile GPU vs CPU (AlexNet) Internally, CNNdroid expresses data parallelism for different layers, instead of leaving to the GPU’s hardware scheduler
  • 16. DeepLearningKit Platform : iOS, OS X and tvOS (Apple TV) DNN Type : CNNs models trained in Caffe Runs on mobile GPU, uses Metal Pro : Fast, directly ingests Caffe models Con : Unmaintained
  • 17. Caffe Caffe for Android https://github.com/sh1r0/caffe-android-lib Sample app https://github.com/sh1r0/caffe-android-demo Caffe for iOS : https://github.com/aleph7/caffe Sample app https://github.com/noradaiko/caffe-ios-sample Pro : Usually couple of lines to port a pretrained model to mobile CPU Con : Unmaintained
  • 18. Running pre-trained models on mobile Mobile Library Platform GPU DNN Architecture Supported Trained Models Supported Tensorflow iOS/Android Yes CNN,RNN,LSTM, etc Tensorflow CNNDroid Android Yes CNN Caffe, Torch, Theano DeepLearningKit iOS Yes CNN Caffe MXNet iOS/Android No CNN,RNN,LSTM, etc MXNet Caffe iOS/Android No CNN Caffe Torch iOS/Android No CNN,RNN,LSTM, etc Torch
  • 19. Building a DL App in 1 week
  • 20. Learn Playing an Accordion 3 months
  • 21. Learn Playing an Accordion 3 months Knows Piano Fine Tune Skills 1 week
  • 22. I got a dataset, Now What? Step 1 : Find a pre-trained model Step 2 : Fine tune a pre-trained model Step 3 : Run using existing frameworks “Don’t Be A Hero” - Andrej Karpathy
  • 23. How to find pretrained models for my task? Search “Model Zoo” Microsoft Cognitive Toolkit (previously called CNTK) – 50 Models Caffe Model Zoo Keras Tensorflow MXNet
  • 24. AlexNet, 2012 (simplified) [Krizhevsky, Sutskever,Hinton’12] Honglak Lee, Roger Grosse, Rajesh Ranganath, and Andrew Ng, “Unsupervised Learning of Hierarchical Representations with Convolutional Deep Belief Networks”, 11 n-dimension Feature representation
  • 25. Deciding how to fine tune Size of New Dataset Similarity to Original Dataset What to do? Large High Fine tune. Small High Don’t Fine Tune, it will overfit. Train linear classifier on CNN Features Small Low Train a classifier from activations in lower layers. Higher layers are dataset specific to older dataset. Large Low Train CNN from scratch http://blog.revolutionanalytics.com/2016/08/deep-learning-part-2.html
  • 26. Deciding when to fine tune Size of New Dataset Similarity to Original Dataset What to do? Large High Fine tune. Small High Don’t Fine Tune, it will overfit. Train linear classifier on CNN Features Small Low Train a classifier from activations in lower layers. Higher layers are dataset specific to older dataset. Large Low Train CNN from scratch http://blog.revolutionanalytics.com/2016/08/deep-learning-part-2.html
  • 27. Deciding when to fine tune Size of New Dataset Similarity to Original Dataset What to do? Large High Fine tune. Small High Don’t Fine Tune, it will overfit. Train linear classifier on CNN Features Small Low Train a classifier from activations in lower layers. Higher layers are dataset specific to older dataset. Large Low Train CNN from scratch http://blog.revolutionanalytics.com/2016/08/deep-learning-part-2.html
  • 28. Deciding when to fine tune Size of New Dataset Similarity to Original Dataset What to do? Large High Fine tune. Small High Don’t Fine Tune, it will overfit. Train linear classifier on CNN Features Small Low Train a classifier from activations in lower layers. Higher layers are dataset specific to older dataset. Large Low Train CNN from scratch http://blog.revolutionanalytics.com/2016/08/deep-learning-part-2.html
  • 29. Building a DL Website in 1 week
  • 30. Less Data + Smaller Networks = Faster browser training
  • 31. Several JavaScript Libraries Run large CNNs • Keras-JS • MXNetJS • CaffeJS Train and Run CNNs • ConvNetJS Train and Run LSTMs • Brain.js • Synaptic.js Train and Run NNs • Mind.js • DN2A
  • 32. ConvNetJS Both Train and Test NNs in browser Train CNNs in browser
  • 33. Keras.js Run Keras models in browser, with GPU support.
  • 34. Brain.JS Train and run NNs in browser Supports Feedforward, RNN, LSTM, GRU No CNNs Demo : http://brainjs.com/ Trained NN to recognize color contrast
  • 35. MXNetJS On Firefox and Microsoft Edge, performance is 8x faster than Chrome. Optimization difference because of ASM.js.
  • 36. Building a DL App in 1 month (and get featured in Apple App store)
  • 37. Response Time Limits – Powers of 10 0.1 second : Reacting instantly 1.0 seconds : User’s flow of thought 10 seconds : Keeping the user’s attention [Miller 1968; Card et al. 1991; Jakob Nielsen 1993]:
  • 38. Apple frameworks for Deep Learning Inference BNNS – Basic Neural Network Subroutine MPS – Metal Performance Shaders
  • 39. Metal Performance Shaders (MPS) Fast, Provides GPU acceleration for inference phase Faster app load times than Tensorflow (Jan 2017) About 1/3rd the run time memory of Tensorflow on Inception-V3 (Jan 2017) ~130 ms on iPhone 7S Plus to run Inception-V3 Cons: • Limited documentation. • No easy way to programmatically port models. • No batch normalization. Solution : Join Conv and BatchNorm weights
  • 40. Putting out more frames than an art gallery
  • 41. Basic Neural Network Subroutines (BNNS) Runs on CPU BNNS is faster for smaller networks than MPS but slower for bigger networks
  • 42. BrainCore NN Framework for iOS Provides LSTMs functionality Fast, uses Metal, runs on iPhone GPU https://github.com/aleph7/braincore
  • 43. Building a DL App in 6 months
  • 44. What you want https://www.flickr.com/photos/kenjonbro/9075514760/ and http://www.newcars.com/land-rover/range-rover-sport/2016 $2000$200,000 What you can afford
  • 45. 11x11 conv, 96, /4, pool/2 5x5 conv, 256, pool/2 3x3 conv, 384 3x3 conv, 384 3x3 conv, 256, pool/2 fc, 4096 fc, 4096 fc, 1000 AlexNet, 8 layers (ILSVRC 2012) Revolution of Depth Kaiming He, Xiangyu Zhang, Shaoqing Ren, & Jian Sun. “Deep Residual Learning for Image Recognition”, 2015
  • 46. 11x11 conv, 96, /4, pool/2 5x5 conv, 256, pool/2 3x3 conv, 384 3x3 conv, 384 3x3 conv, 256, pool/2 fc, 4096 fc, 4096 fc, 1000 AlexNet, 8 layers (ILSVRC 2012) 3x3 conv, 64 3x3 conv, 64, pool/2 3x3 conv, 128 3x3 conv, 128, pool/2 3x3 conv, 256 3x3 conv, 256 3x3 conv, 256 3x3 conv, 256, pool/2 3x3 conv, 512 3x3 conv, 512 3x3 conv, 512 3x3 conv, 512, pool/2 3x3 conv, 512 3x3 conv, 512 3x3 conv, 512 3x3 conv, 512, pool/2 fc, 4096 fc, 4096 fc, 1000 VGG, 19 layers (ILSVRC 2014) input Conv 7x7+ 2(S) MaxPool 3x3+ 2(S) LocalRespNorm Conv 1x1+ 1(V) Conv 3x3+ 1(S) LocalRespNorm MaxPool 3x3+ 2(S) Conv Conv Conv Conv 1x1+ 1(S) 3x3+ 1(S) 5x5+ 1(S) 1x1+ 1(S) Conv Conv MaxPool 1x1+ 1(S) 1x1+ 1(S) 3x3+ 1(S) Dept hConcat Conv Conv Conv Conv 1x1+ 1(S) 3x3+ 1(S) 5x5+ 1(S) 1x1+ 1(S) Conv Conv MaxPool 1x1+ 1(S) 1x1+ 1(S) 3x3+ 1(S) Dept hConcat MaxPool 3x3+ 2(S) Conv Conv Conv Conv 1x1+ 1(S) 3x3+ 1(S) 5x5+ 1(S) 1x1+ 1(S) Conv Conv MaxPool 1x1+ 1(S) 1x1+ 1(S) 3x3+ 1(S) Dept hConcat Conv Conv Conv Conv 1x1+ 1(S) 3x3+ 1(S) 5x5+ 1(S) 1x1+ 1(S) Conv Conv MaxPool 1x1+ 1(S) 1x1+ 1(S) 3x3+ 1(S) AveragePool 5x5+ 3(V) Dept hConcat Conv Conv Conv Conv 1x1+ 1(S) 3x3+ 1(S) 5x5+ 1(S) 1x1+ 1(S) Conv Conv MaxPool 1x1+ 1(S) 1x1+ 1(S) 3x3+ 1(S) Dept hConcat Conv Conv Conv Conv 1x1+ 1(S) 3x3+ 1(S) 5x5+ 1(S) 1x1+ 1(S) Conv Conv MaxPool 1x1+ 1(S) 1x1+ 1(S) 3x3+ 1(S) Dept hConcat Conv Conv Conv Conv 1x1+ 1(S) 3x3+ 1(S) 5x5+ 1(S) 1x1+ 1(S) Conv Conv MaxPool 1x1+ 1(S) 1x1+ 1(S) 3x3+ 1(S) AveragePool 5x5+ 3(V) Dept hConcat MaxPool 3x3+ 2(S) Conv Conv Conv Conv 1x1+ 1(S) 3x3+ 1(S) 5x5+ 1(S) 1x1+ 1(S) Conv Conv MaxPool 1x1+ 1(S) 1x1+ 1(S) 3x3+ 1(S) Dept hConcat Conv Conv Conv Conv 1x1+ 1(S) 3x3+ 1(S) 5x5+ 1(S) 1x1+ 1(S) Conv Conv MaxPool 1x1+ 1(S) 1x1+ 1(S) 3x3+ 1(S) Dept hConcat AveragePool 7x7+ 1(V) FC Conv 1x1+ 1(S) FC FC Soft maxAct ivat ion soft max0 Conv 1x1+ 1(S) FC FC Soft maxAct ivat ion soft max1 Soft maxAct ivat ion soft max2 GoogleNet, 22 layers (ILSVRC 2014) Revolution of Depth Kaiming He, Xiangyu Zhang, Shaoqing Ren, & Jian Sun. “Deep Residual Learning for Image Recognition”, 2015
  • 47. AlexNet, 8 layers (ILSVRC 2012) ResNet, 152 layers (ILSVRC 2015) 3x3 conv, 64 3x3 conv, 64, pool/2 3x3 conv, 128 3x3 conv, 128, pool/2 3x3 conv, 256 3x3 conv, 256 3x3 conv, 256 3x3 conv, 256, pool/2 3x3 conv, 512 3x3 conv, 512 3x3 conv, 512 3x3 conv, 512, pool/2 3x3 conv, 512 3x3 conv, 512 3x3 conv, 512 3x3 conv, 512, pool/2 fc, 4096 fc, 4096 fc, 1000 11x11 conv, 96, /4, pool/2 5x5 conv, 256, pool/2 3x3 conv, 384 3x3 conv, 384 3x3 conv, 256, pool/2 fc, 4096 fc, 4096 fc, 1000 1x1 conv, 64 3x3 conv, 64 1x1 conv, 256 1x1 conv, 64 3x3 conv, 64 1x1 conv, 256 1x1 conv, 64 3x3 conv, 64 1x1 conv, 256 1x2 conv, 128, /2 3x3 conv, 128 1x1 conv, 512 1x1 conv, 128 3x3 conv, 128 1x1 conv, 512 1x1 conv, 128 3x3 conv, 128 1x1 conv, 512 1x1 conv, 128 3x3 conv, 128 1x1 conv, 512 1x1 conv, 128 3x3 conv, 128 1x1 conv, 512 1x1 conv, 128 3x3 conv, 128 1x1 conv, 512 1x1 conv, 128 3x3 conv, 128 1x1 conv, 512 1x1 conv, 128 3x3 conv, 128 1x1 conv, 512 1x1 conv, 256, /2 3x3 conv, 256 1x1 conv, 1024 1x1 conv, 256 3x3 conv, 256 1x1 conv, 1024 1x1 conv, 256 3x3 conv, 256 1x1 conv, 1024 1x1 conv, 256 3x3 conv, 256 1x1 conv, 1024 1x1 conv, 256 3x3 conv, 256 1x1 conv, 1024 1x1 conv, 256 3x3 conv, 256 1x1 conv, 1024 1x1 conv, 256 3x3 conv, 256 1x1 conv, 1024 1x1 conv, 256 3x3 conv, 256 1x1 conv, 1024 1x1 conv, 256 3x3 conv, 256 1x1 conv, 1024 1x1 conv, 256 3x3 conv, 256 1x1 conv, 1024 1x1 conv, 256 3x3 conv, 256 1x1 conv, 1024 1x1 conv, 256 3x3 conv, 256 1x1 conv, 1024 1x1 conv, 256 3x3 conv, 256 1x1 conv, 1024 1x1 conv, 256 3x3 conv, 256 1x1 conv, 1024 1x1 conv, 256 3x3 conv, 256 1x1 conv, 1024 1x1 conv, 256 3x3 conv, 256 1x1 conv, 1024 1x1 conv, 256 3x3 conv, 256 1x1 conv, 1024 1x1 conv, 256 3x3 conv, 256 1x1 conv, 1024 1x1 conv, 256 3x3 conv, 256 1x1 conv, 1024 1x1 conv, 256 3x3 conv, 256 1x1 conv, 1024 1x1 conv, 256 3x3 conv, 256 1x1 conv, 1024 1x1 conv, 256 3x3 conv, 256 1x1 conv, 1024 1x1 conv, 256 3x3 conv, 256 1x1 conv, 1024 1x1 conv, 256 3x3 conv, 256 1x1 conv, 1024 1x1 conv, 256 3x3 conv, 256 1x1 conv, 1024 1x1 conv, 256 3x3 conv, 256 1x1 conv, 1024 1x1 conv, 256 3x3 conv, 256 1x1 conv, 1024 1x1 conv, 256 3x3 conv, 256 1x1 conv, 1024 1x1 conv, 256 3x3 conv, 256 1x1 conv, 1024 1x1 conv, 256 3x3 conv, 256 1x1 conv, 1024 1x1 conv, 256 3x3 conv, 256 1x1 conv, 1024 1x1 conv, 256 3x3 conv, 256 1x1 conv, 1024 1x1 conv, 256 3x3 conv, 256 1x1 conv, 1024 1x1 conv, 256 3x3 conv, 256 1x1 conv, 1024 1x1 conv, 256 3x3 conv, 256 1x1 conv, 1024 1x1 conv, 256 3x3 conv, 256 1x1 conv, 1024 1x1 conv, 512, /2 3x3 conv, 512 1x1 conv, 2048 1x1 conv, 512 3x3 conv, 512 1x1 conv, 2048 1x1 conv, 512 3x3 conv, 512 1x1 conv, 2048 ave pool, fc 1000 7x7 conv, 64, /2, pool/2 VGG, 19 layers (ILSVRC 2014) Revolution of Depth Kaiming He, Xiangyu Zhang, Shaoqing Ren, & Jian Sun. “Deep Residual Learning for Image Recognition”, 2015 Ultra deep
  • 48. ResNet, 152 layers 1x1 conv, 64 3x3 conv, 64 1x1 conv, 256 1x1 conv, 64 3x3 conv, 64 1x1 conv, 256 1x1 conv, 64 3x3 conv, 64 1x1 conv, 256 1x2 conv, 128, /2 3x3 conv, 128 1x1 conv, 512 1x1 conv, 128 3x3 conv, 128 1x1 conv, 512 1x1 conv, 128 3x3 conv, 128 1x1 conv, 512 1x1 conv, 128 3x3 conv, 128 1x1 conv, 512 1x1 conv, 128 3x3 conv, 128 1x1 conv, 512 1x1 conv, 128 3x3 conv, 128 1x1 conv, 512 7x7 conv, 64, /2, pool/2 Revolution of Depth Kaiming He, Xiangyu Zhang, Shaoqing Ren, & Jian Sun. “Deep Residual Learning for Image Recognition”, 2015
  • 49. 3.57 6.7 7.3 11.7 16.4 25.8 28.2 ILSVRC'15 ResNet ILSVRC'14 GoogleNet ILSVRC'14 VGG ILSVRC'13 ILSVRC'12 AlexNet ILSVRC'11 ILSVRC'10 ImageNet Classification top-5 error (%) shallow8 layers 19 layers22 layers 152 layers Kaiming He, Xiangyu Zhang, Shaoqing Ren, & Jian Sun. “Deep Residual Learning for Image Recognition”, 2015 8 layers Revolution of Depth
  • 50. Your Budget - Smartphone Floating Point Operations Per Second (2015) http://pages.experts-exchange.com/processing-power-compared/
  • 51. Accuracy vs Operations Per Image Inference Size is proportional to num parameters Alfredo Canziani, Adam Paszke, Eugenio Culurciello, “An Analysis of Deep Neural Network Models for Practical Applications” 2016 552 MB 240 MB What we want
  • 52. Accuracy Per Parameter Alfredo Canziani, Adam Paszke, Eugenio Culurciello, “An Analysis of Deep Neural Network Models for Practical Applications” 2016
  • 53. Pick your DNN Architecture for your mobile architecture Resnet Family Under 150 ms on iPhone 7 using Metal GPU Kaiming He, Xiangyu Zhang, Shaoqing Ren, Jian Sun, "Deep Residual Learning for Image Recognition”, 2015
  • 54. Strategies to make DNNs even more efficient Shallow networks Compressing pre-trained networks Designing compact layers Quantizing parameters Network binarization
  • 55. Pruning Aim : Remove all connections with absolute weights below a threshold Song Han, Jeff Pool, John Tran, William J. Dally, "Learning both Weights and Connections for Efficient Neural Networks", 2015
  • 56. Observation : Most parameters in Fully Connected Layers AlexNet 240 MB VGG-16 552 MB 96% of all parameters 90% of all parameters
  • 57. Pruning gets quickest model compression without accuracy loss AlexNet 240 MB VGG-16 552 MB First layer which directly interacts with image is sensitive and cannot be pruned too much without hurting accuracy
  • 58. Weight Sharing Idea : Cluster weights with similar values together, and store in a dictionary. Codebook Huffman coding HashedNets Simplest implementation: • Round all weights into 256 levels • Tensorflow export script reduces inception zip file from 87 MB to 26 MB with 1% drop in precision
  • 59. Selective training to keep networks shallow Idea : Augment data limited to how your network will be used Example : If making a selfie app, no benefit in rotating training images beyond +-45 degrees. Your phone will anyway rotate. Followed by WordLens / Google Translate Example : Add blur if analyzing mobile phone frames
  • 60. Design consideration for custom architectures – Small Filters Three layers of 3x3 convolutions >> One layer of 7x7 convolution Replace large 5x5, 7x7 convolutions with stacks of 3x3 convolutions Replace NxN convolutions with stack of 1xN and Nx1 Fewer parameters  Less compute  More non-linearity  Better Faster Stronger Andrej Karpathy, CS-231n Notes, Lecture 11
  • 61. SqueezeNet - AlexNet-level accuracy in 0.5 MB SqueezeNet base 4.8 MB SqueezeNet compressed 0.5 MB 80.3% top-5 Accuracy on ImageNet 0.72 GFLOPS/image Fire Block Forrest N. Iandola, Song Han et al, "SqueezeNet: AlexNet-level accuracy with 50x fewer parameters and <0.5MB model size"
  • 62. Reduced precision Reduce precision from 32 bits to <=16 bits or lesser Use stochastic rounding for best results In Practice: • Ristretto + Caffe • Automatic Network quantization • Finds balance between compression rate and accuracy • Apple Metal Performance Shaders automatically quantize to 16 bits • Tensorflow has 8 bit quantization support • Gemmlowp – Low precision matrix multiplication library
  • 63. Binary weighted Networks Idea :Reduce the weights to -1,+1 Speedup : Convolution operation can be approximated by only summation and subtraction Mohammad Rastegari, Vicente Ordonez, Joseph Redmon, Ali Farhadi, “XNOR-Net: ImageNet Classification Using Binary Convolutional Neural Networks”
  • 64. Binary weighted Networks Idea :Reduce the weights to -1,+1 Speedup : Convolution operation can be approximated by only summation and subtraction Mohammad Rastegari, Vicente Ordonez, Joseph Redmon, Ali Farhadi, “XNOR-Net: ImageNet Classification Using Binary Convolutional Neural Networks”
  • 65. Binary weighted Networks Idea :Reduce the weights to -1,+1 Speedup : Convolution operation can be approximated by only summation and subtraction Mohammad Rastegari, Vicente Ordonez, Joseph Redmon, Ali Farhadi, “XNOR-Net: ImageNet Classification Using Binary Convolutional Neural Networks”
  • 66. XNOR-Net Idea :Reduce both weights + inputs to -1,+1 Speedup : Convolution operation can be approximated by XNOR and Bitcount operations Mohammad Rastegari, Vicente Ordonez, Joseph Redmon, Ali Farhadi, “XNOR-Net: ImageNet Classification Using Binary Convolutional Neural Networks”
  • 67. XNOR-Net Idea :Reduce both weights + inputs to -1,+1 Speedup : Convolution operation can be approximated by XNOR and Bitcount operations Mohammad Rastegari, Vicente Ordonez, Joseph Redmon, Ali Farhadi, “XNOR-Net: ImageNet Classification Using Binary Convolutional Neural Networks”
  • 68. XNOR-Net Idea :Reduce both weights + inputs to -1,+1 Speedup : Convolution operation can be approximated by XNOR and Bitcount operations Mohammad Rastegari, Vicente Ordonez, Joseph Redmon, Ali Farhadi, “XNOR-Net: ImageNet Classification Using Binary Convolutional Neural Networks”
  • 69. XNOR-Net on Mobile
  • 70. Building a DL App and get $10 million in funding (or a PhD)
  • 71. Minerva
  • 72. Minerva
  • 73. DeepX Toolkit Nicholas D. Lane et al, “DXTK : Enabling Resource-efficient Deep Learning on Mobile and Embedded Devices with the DeepX Toolkit",2016
  • 74. EIE : Efficient Inference Engine on Compressed DNNs Song Han, Xingyu Liu, Huizi Mao, Jing Pu, Ardavan Pedram, Mark Horowitz, William Dally, "EIE: Efficient Inference Engine on Compressed Deep Neural Network", 2016 189x faster on CPU 13x faster on GPU
  • 75. One Last Question
  • 76. How to access the slides in 1 second Link posted here -> @anirudhkoul

Top DSC Resources

Follow us on Twitter: @DataScienceCtrl | @AnalyticBridge

Views: 5216

Comment

You need to be a member of Data Science Central to add comments!

Join Data Science Central

Follow Us

Videos

  • Add Videos
  • View All

Resources

© 2017   Data Science Central   Powered by

Badges  |  Report an Issue  |  Privacy Policy  |  Terms of Service