Introduction to TensorFlow
20 likes19,099 views
TensorFlow is an open source software library for machine learning developed by Google. It provides primitives for defining functions on tensors and automatically computing their derivatives. TensorFlow represents computations as data flow graphs with nodes representing operations and edges representing tensors. It is widely used for neural networks and deep learning tasks like image classification, language processing, and speech recognition. TensorFlow is portable, scalable, and has a large community and support for deployment compared to other frameworks. It works by constructing a computational graph during modeling, and then executing operations by pushing data through the graph.
Introduction to TensorFlow
- 1. Introduction to TensorFlow Introduction to TensorFlow GDG Cloud Belgium 29/09/2016 Matthias Feys
- 2. Introduction to TensorFlow 2 About myself (Matthias Feys) work at Datatonic: - Big Data (Dataflow/Spark) - Machine Learning (TensorFlow/sklearn) - DataViz (Tableau/Spotfire) Google Qualified Developer Contact me: - @FsMatt - matthias@datatonic.com
- 3. Introduction to TensorFlow 3 1. What is TensorFlow? 2. Why would you use it? 3. How does it work? + Demo 4. CloudML (alpha) discussion Agenda
- 4. PLACE IMAGE HERE 4 Google TensorFlow ● Originally developed by the Google Brain Team within Google's Machine Intelligence research organization ● TensorFlow provides primitives for defining functions on tensors and automatically computing their derivatives. ● An open source software library for numerical computation using data flow graphs TensorFlow
- 5. 5 Tensor? Simply put: Tensors can be viewed as a multidimensional array of numbers. This means that: ● A scalar is a tensor, ● A vector is a tensor, ● A matrix is a tensor ● ...
- 6. 6 Data Flow Graph? ● Computations are represented as graphs: ● Nodes are the operations (ops) ● Edges are the Tensors (multidimensional arrays) ● Typical program consists of 2 phases: ● construction phase: assembling a graph (model) ● execution phase: pushing data through the graph
- 7. 7 Neural Networks? Deep Learning? ● Neural Networks are represented by the lower figure, not the top one.... ● Link: Tinker with a Neural Network in Your Browser
- 8. Presentation title (Go to View > Master to edit) 8 Source: https://www.udacity.com/course/deep-learning--ud730
- 9. Presentation title (Go to View > Master to edit) 9 Source: https://www.udacity.com/course/deep-learning--ud730
- 10. Presentation title (Go to View > Master to edit) 10 Source: https://www.udacity.com/course/deep-learning--ud730
- 11. Presentation title (Go to View > Master to edit) 11 Source: https://www.udacity.com/course/deep-learning--ud730
- 12. Presentation title (Go to View > Master to edit) 12 Source: https://www.udacity.com/course/deep-learning--ud730
- 13. Presentation title (Go to View > Master to edit) 13 Source: https://www.udacity.com/course/deep-learning--ud730
- 14. Presentation title (Go to View > Master to edit) 14 Source: https://www.udacity.com/course/deep-learning--ud730
- 15. Presentation title (Go to View > Master to edit) 15 Source: https://www.udacity.com/course/deep-learning--ud730
- 16. Presentation title (Go to View > Master to edit) 16 Source: https://www.udacity.com/course/deep-learning--ud730
- 17. Introduction to TensorFlow 17 1. What is TensorFlow? 2. Why would you use it? 3. How does it work? + Demo 4. CloudML (alpha) discussion Agenda
- 18. Introduction to TensorFlow 18 Why would you use NN / Deep Learning? ● Neural Networks (NNs) are universal function approximators that work very well with huge datasets ● NNs / deep networks do unsupervised feature learning ● Track record, being SotA in: ○ image classification, ○ language processing, ○ speech recognition, ○ ...
- 19. 19 Why TensorFlow? There are a lot of alternatives: ● Torch ● Caffe ● Theano (Keras, Lasagne) ● CuDNN ● Mxnet ● DSSTNE ● DL4J ● DIANNE ● Etc.
- 20. Introduction to TensorFlow 20 TensorFlow has the largest community Sources: http://deliprao.com/archives/168 http://www.slideshare.net/JenAman/large-scale-deep-learning-wit h-tensorflow
- 21. Introduction to TensorFlow 21 TensorFlow is very portable/scalable Runs on CPUs, GPUs, TPUs over one or more machines, but also on phones(android+iOS) and raspberry pi’s...
- 22. Introduction to TensorFlow 22 TensorFlow is more than an R&D project - Specific functionalities for deployment (TF Serving / CloudML) - Easier/more documentation (for more general public) - Included visualization tool (Tensorboard) - Simplified interfaces like SKFlow
- 23. Introduction to TensorFlow 23 1. What is TensorFlow? 2. Why would you use it? 3. How does it work? + Demo 4. CloudML (alpha) discussion Agenda
- 24. Introduction to TensorFlow 24 How does it work? Number Recognition w TF explained (in notebook) Speech classification (demo) Great starting point: https://github.com/tensorflow/models Tensorboard notebook: here
- 25. Introduction to TensorFlow 25 Do It Yourself! (in Datalab) Do It Yourself: 1) Open Cloud Shell 2) Paste these commands: 3) Enter the returned EXTERNAL-IP+”:8080” in your browser gcloud container clusters create datalab-cluster --machine-type n1-standard-4 --num-nodes 1 --zone europe-west1-d kubectl run datalab --image=gcr.io/cloud-datalab/datalab:mlbeta2 --port=8080 kubectl expose deployment datalab --type="LoadBalancer" kubectl get service datalab
- 26. Introduction to TensorFlow 26 1. What is TensorFlow? 2. Why would you use it? 3. How does it work? + Demo 4. CloudML (alpha) discussion Agenda
- 28. Introduction to TensorFlow 28 - Curated list of TF resources: https://github.com/jtoy/awesome-tensorflow - Models implemented in TF: https://github.com/tensorflow/models - Slides “TF tricks of the trade”: https://drive.google.com/open?id=x_... - Slides “TF and Deep Learning without a PhD”: https://docs.google.com/presentation/d/... - Blogpost “DL with spark and TF”: https://databricks.com/blog/... - The official documentation: https://www.tensorflow.org/versions/r0.10/... Join: https://www.meetup.com/TensorFlow-Belgium Further reading
- 29. Introduction to TensorFlow 29 Thank you